Update on an Observational, Clinically Useful Gait Coordination Measure: The Gait Assessment and Intervention Tool (G.A.I.T.)

With discoveries of brain and spinal cord mechanisms that control gait, and disrupt gait coordination after disease or injury, and that respond to motor training for those with neurological disease or injury, there is greater ability to construct more efficacious gait coordination training paradigms. Therefore, it is critical in these contemporary times, to use the most precise, sensitive, homogeneous (i.e., domain-specific), and comprehensive measures available to assess gait coordination, dyscoordination, and changes in response to treatment. Gait coordination is defined as the simultaneous performance of the spatial and temporal components of gait. While kinematic gait measures are considered the gold standard, the equipment and analysis cost and time preclude their use in most clinics. At the same time, observational gait coordination scales can be considered. Two independent groups identified the Gait Assessment and Intervention Tool (G.A.I.T.) as the most suitable scale for both research and clinical practice, compared to other observational gait scales, since it has been proven to be valid, reliable, sensitive to change, homogeneous, and comprehensive. The G.A.I.T. has shown strong reliability, validity, and sensitive precision for those with stroke or multiple sclerosis (MS). The G.A.I.T. has been translated into four languages (English, Spanish, Taiwanese, and Portuguese (translation is complete, but not yet published)), and is in use in at least 10 countries. As a contribution to the field, and in view of the evidence for continued usefulness and international use for the G.A.I.T. measure, we have provided this update, as well as an open access copy of the measure for use in clinical practice and research, as well as directions for administering the G.A.I.T.


Gait Coordination Importance and Definition
There have been important discoveries of brain and spinal cord mechanisms controlling gait coordination (e.g., [1][2][3][4][5][6][7]). Additionally, there is growing evidence identifying the neural pathologies and resulting impairments underlying gait dyscoordination after stroke and other neurological diagnoses, and their response to treatment (e.g., [8][9][10][11][12]). For stroke, these include description not only of the disruption of neural control, but its manifestation in terms of gait dyscoordination, for example, abnormal co-contractions [13] and interference with the normal coordinated interactions of the mechanical energetics of the lower limbs [14].
Based on the available discoveries, there is a greater ability to construct more efficacious gait coordination training paradigms for those with neurological injury or disease.
With new interventions targeted at improving the neural drive of gait coordination, it is critical in these contemporary times to use the most precise, sensitive, homogeneous (i.e., domain-specific [15]), and comprehensive measures available [16,17] to assess both gait coordination itself, as well as its underlying domains of strength, limb joint coordinated movement, coordination of muscle co-contractions, proprioception, and balance [18][19][20]. Indeed, others have called for and suggested a standardized set of assessment measures [21], but pertaining to gait coordination, this most recent suggestion was limited to only an incomplete gait measure [16,17].
Gait coordination is defined well by Krasovsky and Levin [22], page 213, as follows: "Locomotor coordination is a context-dependent property of the motor system, having both spatial and temporal components. Spatial coordination is the relationship between the position of different body segments or joints, whereas temporal coordination is the relative timing between segment or joint positions throughout the task. These components are never mutually exclusive. . . . . . " Therefore, an example of a gait coordination measure is 'knee flexion angle at toe-off'. In this example, the spatial component is knee flexion angle and the temporal component is the gait event time of toe-off. Without both the spatial and temporal components simultaneously being performed, there is no measure of coordination. The technologybased gold standard of gait coordination measurement and force production includes gait kinematics and gait kinetics, customarily measured using motion capture and force plate systems in well-equipped gait research laboratories [23]. Studies have shown that gait kinematics is an important quantitative measure of gait coordination and is associated with observed abnormal co-contractions after stroke, causing gait dyscoordination [24].

Observational Gait Coordination Scale with Precision, Sensitivity, Reliability, Validity, Homogeneity, and Comprehensiveness
In clinical practice and in some gait research settings, there is no access to such technology. Furthermore, the time constraints of clinical practice make it impractical to utilize this instrumented technology in its current form due to the labor-intensive process of collecting and analyzing the data. However, observational gait coordination measures can be used in that case. The ideal observational gait coordination measure would be constructed according to the following characteristics: Homogenous, that is, all items measuring gait coordination/dyscoordination, according to the above definition of coordination, and no items assessing compensatory strategies.
As comprehensive as possible, that is, measuring as many coordination joint movements and other gait components as possible.
Numerical scoring scheme overall, and per item, with score numbers within an item assigned across levels of dyscoordination.
Good psychometrics (e.g., reliability, validity). Good sensitivity in measuring the recovery of gait coordination, both endogenous and in response to treatment.
Practical, with a reasonable scoring time and the use of equipment available in most clinical settings.
Two separate research groups conducted a review of observational gait measures [16,17]. They concluded that the Gait Assessment and Intervention Tool (G.A.I.T.) was the most suitable scale for both research and clinical practice compared to other observational gait scales, since it has been proven to be valid, reliable, sensitive to change, homogeneous, and comprehensive (Tables 1 and 2). Since that time, the developers of the G.A.I.T. have received requests and notices indicating that clinicians and researchers are using the G.A.I.T. in the following countries: Turkey, Slovenia, Italy (three separate users), Korea, India (two separate users), Australia (two separate users), Columbia, Spain, Thailand, and the U.S. Additionally, the G.A.I.T. has been translated into the following languages: Spanish [25], Taiwanese [26], and Portuguese (translation is complete, but not yet published). _____ C. Knee flexion moving to extension 0 = normal (knee in 4 • flexion at heel strike, increasing to 15 • flexion at 14% of gait cycle). 1 = normal knee flexion during early midstance phase, then knee extends to neutral. 2 = knee flexion during early midstance phase, then knee extends to full extension range (neutral or beyond) in uncontrolled manner, but not snapping back. 3 = knee in flexion during early midstance phase, then knee abruptly and forcefully extends into end range in an uncontrolled manner. D. Knee extension moving to flexion 0 = normal (knee in 4 • flexion at heel strike, increasing to 15 • flexion at 14% of gait cycle). 1 = knee remains in extension in early midstance, then knee flexes late, but retains control. 2 = knee remains in extension in early midstance, then knee flexes, losing control and regaining control. 3 = knee remains in extension in early midstance, then knee buckles with failure to regain control and requires use of compensatory strategies.

_____
17. Plantarflexion during terminal stance/pre-swing (heel-rise to toe-off) (sagittal plane) (lateral view) 0 = normal (adequate push-off at pre-swing for moving from dorsiflexion position to 10 • plantarflexion. 1 = partial/weak push-off while moving into plantarflexion at toe-off. 2 = absent/lack of plantarflexion; no push-off.    hip neutral with respect to hip abduction/adduction). 1 = hip begins swing in flexion, but reaches normal peak. 1 = > 10 • , but < 30 • hip flexion peak in the sagittal plane. 2 = > 10 • , but < 30 • hip flexion peak, and with hip abduction (e.g., = circumduction). 2 = > 10 • , but < 30 • hip flexion peak, and with hip adduction (e.g., = scissoring). The patient/subject should wear shorts or pants that can be rolled up so that at least the bottom third of his/her thighs are visible. Shirts (upper body clothing) should be tucked into the waistband to ensure viewing of the pelvic position. It is best if the patient/subject wears clothing that is well fitted, not baggy or oversized. If there is little or no (color) contrast between upper and lower body clothing, a gait belt or contrasting band or sash can be placed at the waist. Barefoot ambulation is ideal in order to assess toe position during gait. If this is not deemed safe by the evaluator then the subject/patient should wear his/her regular footwear. It can be helpful to place a piece of contrasting-color tape on each ASIS to help view pelvic movements (this was not used in the published manuscript). • Physical assistance should be minimized since it can affect the patient/subject's gait. If a person walks with the patient/subject without touching him/her, it should be noted as "stand-by assist". Any touching of the patient/subject is considered an assist, even if the person walking with the patient/subject is loosely holding onto a gait belt.

•
The patient/subject should ideally walk without any assistive devices and/or orthoses. If this is not deemed safe by the evaluator, then the patient/subject should use whatever devices necessary to obtain video of his/her gait.

II. Instructions for scoring the Gait Assessment and Intervention Tool (G.A.I.T.)
• Rater training is worthwhile for greatest accuracy in any measure. For the G.A.I.T., one way to conduct rater training for one rater, is to have that rater score a patient's gait pattern on two different days and compare their score on the two different days (intra-rater reliability). Then the rater identifies any discrepancies between their own two rating sessions, analyzes why this occurred, and corrects the thinking or action steps that led to the discrepancy. The steps for training two raters is as follows: each rater scores the same gait pattern; the two raters compare scores for each item; they discuss together how they each arrived at their score; they arrive at a consensus as to how the item should be scored for that gait pattern. If two raters are quite different in their scoring, this process should be repeated using different patients' gait patterns until the raters are in high agreement in scoring. • If more than one evaluator will be scoring the patient's gait, be sure that each evaluator is scoring the same exact gait cycle in the taped video record. We found that stroke survivors can walk with variable gait characteristics across sequential steps in a video record. Therefore, in rater training or reliability testing of the G.A.I.T. in stroke survivors, it is critical to identify the specific gait cycle for a given patient.** • View a middle step of the video record for scoring each item. The first two steps and the last two steps cannot be used for analysis/scoring because they are often affected by the acceleration and deceleration in the gait pattern.

•
For the lateral views, whenever possible use the steps for which the camera is directly opposite the patient/subject. This ensures the best angle for scoring each item. • Some items enable you to input information in addition to entering a score for the item. (for example, indicating the direction of trunk movement, or the specifics of an abnormal shoulder position). These items require a checkmark to be placed on the appropriate line in the form.

•
For items relating to pelvic position (if overhead views are not available), view both the A/P and lateral views in order to gain insight into pelvic movement and position.

•
If an orthotic or supportive device is worn that affects joint movement, the score for the related item would be the midpoint of the abnormal scores for said item. Example: a patient with an AFO receives a score of 2 for item #16.

•
If an assistive device is used for ambulation (cane, walker, etc.), a normal score cannot be given for weight shifting (item 7) or for Trendelenberg (item 8). A 'minimum abnormal' score of 1 must be used.

•
If the patient/subject wears shoes for the assessment and toe position cannot be evaluated, then the items pertaining to toe position should not be scored and the Total possible score adjusted.

•
If minimal physical assistance is provided by one therapist, the scores for items pertaining to trunk alignment/posture and weight shifting should be, at a minimum, the midpoint of the abnormal scores for each item; a higher (more abnormal) score may be indicated. If, however, the physical assistance provided by one therapist appears to be moderate to extensive, or if assistance is provided by more than one therapist, the patient/subject would receive the highest abnormal score. Example: moderate assist of one or minimal assist of 2 people would warrant a score = 3 for item #4; or a score = 2 for item #5, etc.). • If there is anything abnormal about the performance of the item (that is not listed), the patient/subject cannot receive a "0" (normal score) for that item. The evaluator must give a score that he/she judges appropriate based on the abnormality and the other scoring choices offered for the item. • Comments pertaining to abnormalities, deviations, and/or compensations not listed on the G.A.I.T. form should be mentioned in the comment section. • A total score of zero for the comprehensive form = totally normal gait (i.e., no abnormalities). The lower the overall score = the more normal the gait. It may be instructive to score both the patient/subject's extremities for a more accurate accounting of the gait pattern. For the G.A.I.T., each side should be score separately. Alternatively, if using a smartphone camera and its stop-frame capability, eliminate the transparency step and simply place a small goniometer on the phone screen, as described above, to measure the joint angle.

Use of G.A.I.T. for Those with Stroke
The G.A.I.T. psychometrics were originally tested in stroke survivors. The G.A.I.T. was constructed with items measuring the spatial coordinated movement components of gait (movement excursion), that occur at specified temporal events of stance and swing phases of the gait cycle. Within each item of the measure, deviation from normal coordination is scored at multiple levels of dyscoordination. Psychometrics were studied and found to be good, according to intra-rater reliability (intraclass correlation (ICC) = 0.98; p = 0.0001; 95% confidence interval (CI)= 0.95-0.99), and inter-rater reliability (ICC = 0.83; p = 0.007; 95% CI = 0.32-0.96), including between an experienced and an inexperienced clinician (ICC = 0.996; p = 0.0001; 95% CI = 0.986-0.999) [27]. The G.A.I.T. showed measurement sensitivity in identifying a difference between two treatment groups in the recovery of gait coordination from pre-to post-treatment (that is, the G.A.I.T. showed that group 1 had a significant additive effect (parameter statistic = 1.10, p = 0.045, 95% CI= 0.023-2.18)) [28].

MCID
Very recently, two independent groups have estimated the minimal clinically important difference (MCID) for the G.A.I.T. in stroke survivors [29,30]. In a group of subacute patients (mean time since stroke was 45 days), a change in the G.A.I.T. score between 1.5 and 4 points was deemed the MCID value [29]. In a study of chronic stroke, the G.A.I.T. scores were well correlated to the Functional Ambulation Category (FAC) measure (r = 0.73; [30]. In that study, the G.A.I.T. was studied for MCID according to two anchors, the FAC and the speed-based functional categories of household, limited community, or full community ambulation [31]. They proposed the following G.A.I.T. MCID value: 11.8 points related to FAC (Functional Ambulation Category) level 3 or household ambulator gait classification; and 5.19 points related to FAC levels 4 and 5, or limited community ambulator, or full community ambulator classification [30]. In this particular study [30], the G.A.I.T. was scored bilaterally and averaged, which is a novel approach to the utilization of the measure; this procedure may have influenced the value of the MCID by diluting (averaging out) any improvement in the paretic limb and thus resulting in a higher calculated MCID. Given its advantageous characteristics [16,17], additional analyses to interpret the MCID value levels for meaningful change according to the G.A.I.T. are warranted; at the same time, these current studies are an important contribution to the field.

Use of the G.A.I.T. for Those with MS
Recently, the G.A.I.T. psychometrics were studied for those with multiple sclerosis (MS). Construct validity was tested [32]. The results showed a high construct validity, with correlations between the G.A.I.T. and the Rivermeade Gait Assessment (RVGA) at r > 0.90, and ranging in correlation with the Tinetti Gait Scale (TGS) from −0.62 to −0.59. We should note, in this regard, that in prior work, the authors of [17] compared the G.A.I.T. versus other gait measures such as the RVGA and TGS, and found the G.A.I.T. to be the most suitable scale for stroke survivors for both research and clinical practice compared to the other observational gait scales, since it has been proven to be valid, reliable, sensitive to change, homogeneous, and comprehensive. Further, for the MS population, they found that correlations were lower for scales that included speed, such as the Timed Up and Go Test. Reliability of the G.A.I.T. was high for use with those with MS (intraclass correlation coefficient for the intra-rater reliability, e.g., r = 0.91; 95% CI 0.85-0.95 for the right side; and inter-rater reliability, e.g., r = 0.91 (95% CI 0.85-0.95) for the right side) [33].

Contribution to the Field
In view of the evidence for the continued usefulness and international use for the G.A.I.T. measure, we have provided this update, as well as an open access copy of the measure for use in clinical practice and research (Table 1). We are also providing the prior published directions for administering the G.A.I.T., enhanced with additional information obtained from its continued use (Table 2). Institutional Review Board Statement: No subjects were studied for this report.