Is the Dynamic Gait Index a useful outcome to measure balance and ambulation in patients with cerebellar ataxia?

Abstract

Background:

Ataxia can adversely affect balance and gait and increase the incidence of falls, which puts individuals at greater risk for injury. Thus, interventions focused on balance and gait are integral in rehabilitation training. In order to determine if rehabilitation interventions are effective, we need an outcome measure to detect change. To our knowledge, no activity level outcome measures have been established for balance and gait in cerebellar ataxia.

Objective:

The aim of the current study is to determine the reliability and validity of the Dynamic Gait Index (DGI) for ataxia.

Design:

Twenty adult participants (23-84 years) with ataxia were evaluated to assess construct validity, inter-rater reliability, and same day test-retest reliability of the DGI.

Methods:

Participants completed ataxia-specific impairment level outcome measures, as well as the DGI. In addition to the in-person rater, three additional physical therapists scored video recordings of DGI test and retests. Construct validity was assessed via Spearman’s rank order correlation coefficient (Spearman’s rho) between the impairment measures (Scale for Assessment and Rating of Ataxia (SARA), International Cooperative of Ataxia Rating Scale (ICARS)) and the DGI. Reliability was assessed by Spearman’s rho and Intraclass Correlation Coefficient ICC (2,1).

Results:

In terms of construct validity, we found significant correlations between the activity level DGI and impairment level outcome measures (−0.81 for SARA; −0.88 with ICARS). The interrater reliability of the DGI applied to participants with ataxia was high (Spearman rho: range 0.71-0.98; ICC (2,1) 0.98) as was test-retest reliability (Spearman rho: 0.95; ICC (2,1) 0.98).

Conclusion:

We showed that the DGI is a reliable and valid outcome measure to be used in the clinic for individuals with cerebellar ataxia. The DGI had excellent inter-rater and test-retest reliability for raters with varying years of clinical experience. Therefore, the DGI can be a useful clinical outcome measure for assessing balance and ambulation for individuals with cerebellar ataxia.

Introduction:

Cerebellar ataxia can result from an assortment of heath conditions including hereditary syndromes and acquired injuries.^1-2 Prevalence of hereditary ataxia has been found to be 5.6/1,000,000,³ whereas 33% of individuals with traumatic brain injuries⁴ and 80% of individuals with multiple sclerosis are ataxic.⁵ Prevalence of acquired ataxia has been found to be 27/1,000,000.⁶ Regardless of the origin, clinicians assess motor presentations the same way. Therefore, we need a measure that is useful for a variety of etiologies.

Ataxia can adversely affect both balance and gait. Individuals with cerebellar ataxia exhibit a characteristic gait pattern of truncal instability, variable foot placement, inconsistent step length, and path veering.^{7, 8} These balance and gait impairments often lead to dependence for activities of daily living, increase the incidence of falls, and reduce activity and community participation^9-12. Therefore, rehabilitation interventions focused on balance and gait are integral in training for cerebellar ataxia.

To ensure the success of rehabilitation plans of care, therapists use outcome measures to assess patient progression. Two outcome measures, the International Cooperative Ataxia Rating Scale (ICARS) and the Scale for Assessment and Rating of Ataxia (SARA), are standard outcome measures used by clinicians to assess the degree of ataxia.^{13, 14} The ICARS and the SARA have excellent inter-rater reliability.^13-16 However, these outcome measures evaluate movement impairments and do not assess balance and gait at the activity level of the International Classification of Functioning (ICF). The ICF classifies impairments as deviations in body function. Whereas activity is defined as the completion of a task.^{17, 18} We recently studied an activity level outcome measure, the Action Research Arm Test (ARAT), and showed it to be reliable and valid for individuals with ataxia.²⁰ To our knowledge, only the ARAT has been validated for activity level for arm and hand control,²⁰ and no activity level outcome measures have been established for balance and gait in cerebellar ataxia.

The Dynamic Gait Index (DGI) is an activity level outcome measure that has shown strong psychometric properties for balance and gait in non-ataxic patient populations (e.g. Multiple Sclerosis, Parkinson’s Disease, stroke, and vestibular disorders).^21-32 In these populations, the DGI has been found to have excellent inter-rater reliability,^{21, 32} criterion validity,^{22-24, 28} test-retest^{23-27, 31} and inter-rater reliability.^{24, 29, 30} Thus, the DGI may be a useful outcome measure for assessing balance and gait in individuals with ataxia.

The aim of the current study is to determine the reliability and validity of the Dynamic Gait Index (DGI) for individuals with ataxia. We hypothesize that the DGI will be both a reliable and a valid tool for use in the clinic. If our hypothesis is correct, the DGI could serve as an activity level outcome measure that could be used to objectively evaluate rehabilitation interventions for individuals with ataxia.

Methods:

Participants:

Twenty adult participants (age 23-84 years) with ataxia were recruited from the Johns Hopkins Hospital Ataxia Clinic. See Table 1 for background characteristics. All participants met inclusion criteria of cerebellar damage from stroke, tumor, degeneration or presence of clinical ataxia. Exclusion criteria included: 1) presence of extrapyramidal clinical signs, 2) unstable vitals, such as uncontrolled hypertension (>190/110 mmHg), 3) dementia, 4) peripheral vestibular loss, 5) severe aphasia, 6) orthopedic conditions that impaired lower extremity function, 7) pregnancy. Of note, all participants had intact strength and sensation, were ambulatory without the use of assistive devices, and reported a history of falls in the past year. All participants provided written consent per the Johns Hopkins School of Medicine Institutional Review Board (IRB).

Table 1.

Demographic Data and Outcome Measure Scores

Participant	Sex (M/F)	Age (years)	Diagnosis	ICARS Total	ICARS Posture & Gait	SARA	DGI (Test)	DGI (Re- Test)
1	M	75	A	32	10	14	15	-
2	F	54	SCA 7	46	22	18	0	5
3	F	33	SCA 1	30	10	8	13	-
4	F	73	A	24	8	8	12	-
5	F	70	SCA 3	24	6	7.5	13	13
6	M	58	A	49	12	-	2	7
7	M	35	Tu	48	18	18	6	3
8	M	23	A	37	13	16	7	7
9	M	47	A	21	8	7.5	14	14
10	M	84	AI	10	6	4	21	21
11	F	24	A	4	0	2	21	21
12	M	49	A	27	7	8	15	15
13	M	55	A	40	12	11	12	10
14	M	55	A	57	17	24	0	-
15	M	57	SCA 17	18	6	8	17	16
16	F	65	A	24	6	7.5	17	16
17	F	52	SCA 14	18	5	6.5	19	22
18	M	54	SCA 1	38	15	14	9	9
19	M	73	SCA 2	34	15	14	14	14
20	M	69	A	30	12	10.5	15	15
Group	Males: n=13	Mean:55.25	-	Mean:30.55	Mean:10.4	Mean:10.87	Mean:12.10	Mean:13.00
	Females: n=7	SD:16.86	-	SD:13.50	SD:5.29	SD:5.49	SD:6.30	SD:5.76

Demographic data for individual subjects. M: male, F: female; SCA: spinocerebellar ataxia; AI: autoimmune ataxia; Tu: tumor; A: ataxia of unknown origin; SARA: Scale for Assessment and Rating of Ataxia; ICARS: International Cooperative Ataxia Rating Scale; DGI: Dynamic Gait Index. SARA, ICARS & DGI (Test) n=20. DGI (Re-Test) n=16.

Outcome Measures:

Scale for Assessment and Rating of Ataxia (SARA):

The SARA is an 8-item measure that contains sitting and standing balance, gait, and limb kinetic functions. The maximum score for the SARA is 40, which indicates severe ataxia. All items were performed as described by the measure¹⁴ and took approximately 10 minutes to perform.

International Cooperative of Ataxia Rating Scale (ICARS):

The ICARS consists of 19-items and is divided into four subscales: posture and gait disturbance, kinetic function, speech disorder, and oculomotor disorders. The total possible ICARS score is 100, with higher scores indicating greater impairments. All items were performed as described by the tool¹⁴ and took approximately 20 minutes to perform.

We included both the SARA and the ICARS because both measures are frequently used in clinical practice and are reported in clinical research findings. There are no established thresholds or cutoff scores to indicate mild, moderate, or severe disease level.

Dynamic Gait Index (DGI):

The DGI is an eight-item outcome measure for the assessment of balance, functional mobility, and gait. We note that the clinic in which participant evaluations took place did not have access to stairs. As a result, the stair climbing item of the DGI was scored by subjective patient report to allow for completion of the measure. Additionally, it has been shown that the removal of the stair portion of the DGI does not affect the validity or the integrity of the tool.³³ Each item of the DGI is scored on a 4-point scale. A score of a 3 indicates no gait dysfunction, a score of a 2 indicates minimal dysfunction, a score of a 1 represents moderate dysfunction, and a scores of a 0 denotes severe dysfunction. Maximum score is 24.³⁴ The test takes approximately 10 minutes to administer.

Study Design:

All data were collected in the clinical setting, in a medical home model. This model involves individuals coming to the clinic for a day of comprehensive assessments by each of their medical providers (physicians, genetic councilors, social workers, physical therapists, occupational therapists, speech therapists, and laboratory work up). Participants received seated rest breaks of approximately 15 minutes between providers to reduce fatigue that may be experienced during this long day of clinical care. Subjects participated in a single, 90-minute physical therapy testing session during which they completed a neurological examination, the ICARS, the SARA and two repetitions of the DGI: one initial assessment and one re-test assessment. All 20 participants recruited for the study completed the ICARS and the initial DGI assessment. Due to time constraints (late arrival to the clinic and/or increased time spent in prior medical home model visit), one subject was unable to complete the SARA and 4 participants were unable to complete the re-test DGI assessment. For subjects who completed both the initial and re-test DGI assessments, the inter-test interval was approximately 60 minutes. This interval was selected as it was not a long enough delay to be impacted by fatigue or time of day effects that could influence motor performance. Additionally, since people with ataxia have movement inconsistencies, we wanted to see if their performance in activities would also be variable within this short time frame. Note that the sample studied here is distinct from the participants in our recent study of the ARAT.²⁰

A total of four raters assessed the DGI test and DGI re-test. All outcome measures were administered and scored in person by one of two licensed PTs. One PT, rater A, had >5 years of clinical experience and completed 15/20 of the in-person assessments. The second PT, rater B, had >26 years of clinical experience and completed 5/20 of the in-person evaluations.

All DGI assessments were video recorded. Videos were uploaded to a secured drive and independently scored by two additional raters (PTs with clinical experience of 8 and 22 years. Additionally, raters A and B scored the videos of those participants that they did not assess in-person. Prior to scoring, all raters underwent one training session on standard performance of the DGI and were given written instructions, as provided by the instrument, detailing performance, and scoring of each item. Raters were blinded to the scores of other raters. All raters scored the test and re-test DGI assessments for each participant. Inter-rater reliability was assessed at both test and re-test because it was uncertain if fatigue during re-test would impact movement variability and the raters’ reliability. Additionally, the neurological examination was performed to ensure that the participants’ movement dysfunction was due to cerebellar dysfunction, and not strength, sensory, tone or extrapyramidal impairments.

Data Analysis:

Spearman’s rho, α=0.05, was used to analyze construct validity, inter-rater reliability, and test-retest reliability due to the non-continuous ordinal nature of the data. Previous studies have also combined ordinal scores to utilize ICCs to assess interrater reliability;^35-37 therefore, an ICC (2,1) was evaluated, α<0.05. In-person scores given by a single rater were used in the analysis of correlations between outcome measures and the correlations between DGI items (Tables 1, 3 and 6). Video recorded data that were scored by a total of 4 raters was used to analyze inter-rater reliability and test-retest reliability (Tables 2, 4, and 5). Correlations with the SARA only included the participants who completed the SARA. Likewise, test-retest correlations were performed with the participants who performed the re-test. Statistical Package for the Social Sciences (SPSS) was used for all data analysis. Consistent with established interpretations, correlations <0.3 were considered small relationships, 0.3-0.7 moderate relationships, and >0.7 large relationships.³⁸

Table 3.

Correlations between outcome measures

	ICARS	SARA	DGI (Test)	DGI (Re-Test)
ICARS	1.0	-	-	-
SARA	0.94**	1.0	-	-
DGI (Test)	−0.88**	−0.81**	1.0	-
DGI (Re-Test)	−0.84**	−0.78**	0.95**	1.0

Spearman’s rho Inter-rater reliability; ICARS: International Cooperative Ataxic Rating Scale, n=20; SARA: Scale for Assessment and Rating of Ataxia n=19; DGI: Dynamic Gait Index Test, n=20; Dynamic Gait Index Re-Test; n=16

^**indicates significance at the 0.01 level (2-tailed).

Table 6.

Construct Validity: Individual DGI Item Comparison with Impairment Outcome Measures

	ICARS	ICARS P & G	SARA	Item 1	Item 2	Item 3	Item 4	Item 5	Item 6	Item 7
ICARS	1.0	-	-	-	-	-	-	-	-	-
SARA	0.94**	-	1.0	-	-	-	-	-	-	-
Item 1	−0.87**	−0.83**	−0.87**	1.0	-	-	-	-	-	-
Item 2	−0.77**	−0.74**	−0.79**	0.83**	1.0	-	-	-	-	-
Item 3	−0.71**	−0.57**	−0.62**	0.74**	0.70**	1.0	-	-	-	-
Item 4	−0.77**	−0.68**	−0.64**	0.59**	0.63**	0.67**	1.0	-	-	-
Item 5	−0.82**	−0.70**	−0.78**	0.77**	0.72**	0.76**	0.79**	1.0	−	-
Item 6	−0.82**	−0.79**	−0.77**	0.90**	0.83**	0.72**	0.60**	0.66**	1.0	-
Item 7	−0.78**	−0.76**	−0.70**	0.85**	0.84**	0.67**	0.72**	0.69**	0.89**	1.0

Spearman’s rho Inter-rater reliability for the individual test items of the DGI: Dynamic Gait index (n=20); the SARA: Scale for Assessment and Rating of Ataxia (n=19); and the ICARS: International Cooperative of Ataxia Rating Scale (n=20); ICARS P & G: International Cooperative of Ataxia Rating Scale Posture and Gait Subscales (n=20)

^**indicates significance at the 0.01 level (2-tailed).

Table 2.

Inter-Rater Reliability of the DGI Test and DGI Re-Test

	Rater 1 (Test)	Rater 2 (Test)	Rater 3 (Test)	Rater 4 (Test)	Rater 1 (Re- Test)	Rater 2 (Re- Test)	Rater 3 (Re- Test)	Rater 4 (Re- Test)
Rater 1 (Test)	1.0	-	-	-	-	-	-	-
Rater 2 (Test)	0.80**	1.0	-	-	-	-	-	-
Rater 3 (Test)	0.95**	0.89**	1.0	-	-	-	-	-
Rater 4 (Test)	0.88**	0.85**	0.87**	1.0	-	-	-	-
Rater 1 (Re-Test)	0.97**	0.71**	0.91**	0.86**	1.0	-	-	-
Rater 2 (Re-Test)	0.78**	0.89**	0.84**	0.75**	0.73**	1.0	-	-
Rater 3 (Re-Test)	0.94**	0.83**	0.98**	0.87**	0.90**	0.85**	1.0	-
Rater 4 (Re-Test)	0.88**	0.86**	0.87**	0.96**	0.84**	0.78**	0.88**	1.0

Spearman’s rho Inter-rater reliability for the Dynamic Gait Index Test n=20 and Dynamic Gait Index Re-Test; n=16

^**indicates significance at the 0.01 level (2-tailed).

Table 4.

Intraclass Correlation Coefficient ICC (2,1) DGI Test-Retest

	ICC	95% CI LB	95% CI UB	df
Dynamic Gait Index	0.98	0.96	0.99	15

ICC: Intraclass Correlation Coefficient; CI: Confidence Interval; LB: lower bound; UB: upper bound; df: degrees of freedom; DGI: Dynamic Gait Index Test, Test n=20, Re-test n=16. Reliability for all four raters scores for DGI Test and DGI Re-Test.

Table 5.

Inter-Rater Reliability, Intraclass Correlation Coefficient ICC (2,1) for individual DGI Items

	ICC	95% CI LB	95% CI UB	df
Item 1	0.94	0.87	0.97	19
Item 2	0.93	0.88	0.97	19
Item 3	0.89	0.78	0.95	19
Item 4	0.94	0.87	0.97	19
Item 5	0.89	0.78	0.95	19
Item 6	0.90	0.81	0.85	19
Item 7	0.92	0.84	0.96	19

ICC: Intraclass Correlation Coefficient; CI: Confidence Interval; LB: lower bound; UB: upper bound; df: degrees of freedom; DGI: Dynamic Gait Index Test; n=20

Results:

Construct Validity:

We found that the average difference in scores between test and retest was generally small, but there were two individuals (subject 2, subject 6) whose scores differed by 5 points. Table 3 shows the correlations between the impairment level outcome measures (SARA and ICARS), and the activity level outcome measure (DGI). As expected, the ataxia-specific impairment level measures were highly correlated with one another (Spearman’s rho at the two tailed, α<0.05, ICARS and SARA of 0.94). Importantly, we also found that the activity level measure of DGI was strongly correlated with each of the ataxia-specific impairment level measures (Spearman’s rho at the two tailed, α<0.05: ICARS and DGI Test (−0.88), ICARS and DGI Re-Test (−0.84), SARA and DGI Test (−0.81), SARA and DGI Re-Test (−0.78), and DGI Test and DGI Re-Test (0.95)).

Inter-rater reliability

All four raters showed excellent inter-rater reliability. A large, significant relationship was found between all raters for the initial DGI and re-test DGI assessments (Spearman’s rho range of 0.71-0.98; Table 2). The Intraclass Correlation Coefficient ICC (2,1) also showed excellent inter-rater reliability for the initial and re-test DGI assessment scores (0.98, See Table 4). To determine whether there was a difference in inter-rater reliability for individual DGI items, a post hoc assessment of was performed for the seven DGI items scored by video. The inter-rater reliability was excellent for all items of the DGI with ICC (2,1) ranging from 0.78-0.97 (Table 5).

Test-retest reliability

Given the high amount of movement variability that is typical in this population, we were not sure how similar an individual’s scores would be from test to re-test, even with just a 60-minute gap. However, we found large, significant correlations between the test and re-test DGI assessments (Table 3: Spearman’s rho = 0.95; Table 4: ICC (2,1) = 0.98).

Individual Item Comparison

We were also interested in whether certain items on the DGI were more or less correlated with ataxia-specific impairments than others. Therefore, we performed a set of post-hoc analyses. We found that all DGI items had significant, large, negative correlations with the ICARS (Item 1 −0.87, Item 2 −0.77, Item 3 −0.71, Item 4 −0.77, Item 5 −0.82, Item 6 −0.82, Item 7 −0.78) and the SARA (Item 1 −0.87, Item 2 −0.79, Item 3 −0.62, Item 4 −0.64, Item 5 −0.78, Item 6 −0.77, Item 7 −0.70). See Table 6.

Of note, using this data set, we ran a post hoc power analysis and determined based off this data a sample size of 15 would have been sufficient for a power of 0.8.

Discussion:

The aim of the current study was to assess the reliability and validity of the DGI for persons with ataxia. Our results showed that the DGI had excellent construct validity when compared to standard ataxia impairment-level scales, the ICARS and the SARA. That is to say that the more impairment detected by the total scores on the ICARS and SARA, the lower the function detected by the total score on the DGI. Moreover, each individual item of the DGI showed large negative correlations with the ICARS and SARA total scores. These results indicate that the DGI successfully captures the dysfunctional characteristics of ataxic gait and does so equally well across all individual items. The presence of strong negative relationships highlights the link between impairment and activity level measures – it can be speculated that as an individual’s impairment increases (higher scores on the ICARS and SARA), their activity levels will decrease (lower scores on the DGI).

One strength of this study is that we included a heterogenous group of cerebellar dysfunction diagnoses, which resulted in a wide range of impairment levels (ICARS range 4-57; SARA range 2-18). The strong correlations found between these two impairment measures and the DGI suggests that the DGI can capture balance and gait deficits in individuals with varying degrees of ataxia-based impairments. Additionally, the participants presented with a variety of ataxia etiologies including: genetic degenerative cerebellar disease, acquired cerebellar insult, autoimmune conditions. While the sample size is small, this variability in etiology is representative of the mixed patient populations treated in rehabilitation settings. There was also a wide age range to the participants, 24-84 years old. Although we were not powered to split the participants into subgroups based on age, observation of the scores does not indicate a performance trend based on age. Together, these results support the generalizability of the study results to the clinical setting.

Our work suggests that the DGI might have wide ranging clinical utility for people with ataxia. This is supported by the fact that its psychometric properties were so high even with a wide range of impairment levels and etiologies, and by the fact that those administering the measure had wide ranges of clinical/ataxia/DGI-administration experience. Inter-rater reliability was found to be excellent between four raters of varying years of PT experience (range: 5-26 years). This high reliability was identified for the total DGI Test score, the individual item scores, and the total DGI Re-Test score. Given the inconsistent nature of ataxic gait, the high inter-rater reliability is particularly important, as it indicates that clinicians with varying levels of experience are able to consistently identify the ataxic gait characteristics of individuals with ataxia caused by diverse medical conditions. Depending on the cause of ataxia, an individual may be ambulatory for life or the individual may regress to become wheelchair dependent. Thus, the DGI may be a useful tool to assess dynamic balance and ambulation for individuals with non-progressive ataxia throughout their rehabilitation tenure and may be a useful outcome measure for individuals with progressive ataxia until the point of which they are no longer ambulatory. Therefore, the results of this study suggest that the DGI can be utilized in the clinic to assess dynamic balance and gait for individuals with ataxia from varying etiologies.

Another frequent barrier to outcome measure application is the stability of the measure over multiple applications. Moreover, ataxic gait is notoriously inconsistent even within a single day.^{7, 8} Therefore, it was particularly important to determine whether DGI scores were reliable within a session for this patient population. Importantly, we found that the test-retest reliability for the DGI administered to participants with ataxia was excellent.

Future Directions/Study Limitations:

One limitation of the study relates to the selection of the outcome measure. The DGI is a dynamic balance and ambulation outcome measure and is not appropriate for individuals who are non-ambulatory. Thus, this measure may have limited utility for individuals with degenerative forms of ataxia that lead to a reduction in mobility over time.

Additionally, our test-retest intertrial interval was short, approximately 60 minutes. While this interval captures the nature of the participants’ current balance and gait presentation, it does not represent the change in movement pattern that may be observed over time or following a rehabilitation intervention. This short time interval could have the potential to promote a training or learning effect for the participants. However, we did not see a systematic increase in performance from test to re-test (only 3/16 participants increased) nor a systematic decrease in performance (only 4/16 participants decreased) to suggested fatigue. While unlikely, given the short time interval, there is potential that the raters may have been able to recall their test scores during their re-test evaluation. Future studies would benefit from an increased time interval to address this concern. Of note, two participants had an increase in their test retest scores. For these participants, post hoc analysis did not show an increase in one specific item of the DGI, but a small increase across the items. This increase across the task, may indicate improved patient confidence as the session progressed.

Falls were extremely prevalent in our participants with ataxia; 100% of individuals reported at least one fall in the past year. In other neurological patient populations, the DGI has been validated as a means of ascertaining an individual’s fall risk.^{22, 40} The present study was not designed nor powered to predict falls. However, now that we have established the reliability and validity of the DGI for people with ataxia, future work studies should seek to establish falls prediction cutoff scores to further bolster the clinical utility of the measure. Additionally, future studies are needed to address minimal detectible change and minimal clinically important differences for patients with ataxia.

Conclusion:

In people with ataxia, difficulties with gait and dynamic balance dysfunction are prevalent,⁷ have a profound negative impact on the quality of life,⁹ and are central/primary targets of neurorehabilitation intervention. In spite of this, until now there has not been a standardized activity level outcome measure of gait and dynamic balance for this patient population. Here, we demonstrate that the DGI is a valid outcome measure for individuals with cerebellar ataxia, who are ambulatory, regardless of the ataxia etiology. Additionally, the DGI had excellent inter-rater and test-retest reliability for raters with varying years of clinical experience. Therefore, the DGI can be a useful clinical outcome measure for assessing dynamic balance and ambulation for individuals with cerebellar ataxia.

Highlights.

• Rehabilitation interventions for balance and gait are integral in training for ataxia

• The DGI is a reliable and valid measure in the clinic for individuals with ataxia

• DGI has high inter-rater & test-retest reliability for raters with varying experience