Journal of Orthopaedics

Introduction:
Prior to using a given method of assessing alterations in the measured variables, there is a need to evaluate the method's reliability and validity. There is no published data on the GAITRite system's reliability in patients with knee OA The purpose of this study was to examine the reliability of a computerized mat (GAITRite system) in patients with knee osteoarthritis (OA).
Methods:
Forty-one patients with bilateral knee OA participated in the study. Patients were asked to walk barefoot at a self-selected speed on the GAITRite mat. A second test was conducted after a 15 minute period of rest. The GAITRite mat calculated the spatiotemporal parameters of gait in each session. Test-retest reliability was evaluated using Intraclass Correlation Coefficient (ICC) and differences between test scores were evaluated using Wilcoxon ranked test.
Results:
ICCs showed high test-retest reliability in all measured variables (ICC ranged between 0.86 – 0.95). In addition, all spatio-temporal parameters of gait showed similar results in both sessions, however, significant differences between the two tests were found in gait velocity and cadence (p<0.05). These differences were consistent over the entire study group and are thought to be related to an increase in joint stiffness after rest that occurs in patients with knee OA.
Conclusions:
Our results demonstrated high reliability of this gait analysis mat in a population with knee OA.

Introduction
Spatial and temporal gait parameters are often used in gait measurements and are quantified by various methods (1, 2). Prior to using a given method of assessing alterations in the measured variables, there is a need to evaluate the method's reliability and validity in healthy populations as well as in specific pathologic populations. Several studies have tested the validity and reliability of a relatively new computerized mat (GAITRite system) (3) and reported it to be good to excellent in healthy populations (4, 5). Previous studies have reported differences in gait patterns between healthy population and patients with knee osteoarthritis (OA) and also between genders (6, 7), hence it is important to evaluate the validity and reliability of the computerized mat in this specific population. An article published in the late 90’s by Fransen et al. found high reliability of gait measurements in patients with knee OA using electronic footswitches worn by the patients (8). To the best of our knowledge, there is no published data on the GAITRite system's reliability in patients with knee OA. While a preliminary presentation showed high consistency in the walking patterns of adults with knee OA (9), a more thorough examination is necessary.
The aim of this study was to examine the reliability of the GAITRite system in patients with knee OA.

Methods
2.1. Patients
All patients gave written informed consent before entering the study. The protocol was approved by the Institutional Helsinki Committee Registry (Helsinki registration number 185/07, NIH No. NCT00599729). The study was conducted at the Orthopedics Outpatient Clinic of Assaf Harofeh Medical Center in Zerifin, Israel, and from the AposTherapy Center in Herzliya, Israel. Eligibility was defined as symptomatic bilateral medial compartment knee OA for at least six months, fulfillment of the American College of Rheumatology clinical criteria for OA of the knee (10) and radiographic confirmed OA of the knee by the senior author according to the Kellgren & Lawrence (KL) scale (11). Exclusion criteria were acute septic arthritis, inflammatory arthritis, history of knee buckling or recent knee injury, spinal or vascular claudication, use of gait aides, lack of physical or mental ability to perform or comply with the treatment procedure, a history of pathological osteoporotic fracture, and symptomatic degenerative arthritis in lower limb joints other than the knees.
2.2 Apparatus
The gait spatio-temporal parameters were measured using the GAITRite system (Fig. 1). The gait spatio-temporal characteristics were processed and stored by a Hewlett Packard computer using GAITRite Platinum, Version 3.9 software.
2.3. Protocol
All patients underwent anthropometric measurements. Leg length was measured in a standardized method with the patients standing in an upright position. The clinician located the greater trochanter and lateral malleolus on each leg and used measuring tape to measure the distance from the greater trochanter to the floor passing through the lateral malleolus (Fig. 2). All patients walked barefoot at a self-selected speed on the computerized mat. Patients walked three meters before and after the mat to allow sufficient acceleration and deceleration time outside the measurement area. Patients in the current study walked at self-selected speeds since previous works indicated a higher consistency in spatio-temporal parameters at such self-selected walking speeds (12). Following the gait test patients were asked to sit on a standard chair for 15 minutes. A second test was conducted after the 15 minutes of sitting.
The following spatio-temporal parameters were evaluated: absolute velocity (cm/s), normalized velocity (cm/s/leg length), cadence (steps/min), step length (cm), normalized step length (cm/leg length), base of support (cm), swing phase (% gait cycle), stance phase (% gait cycle), single limb support phase (SLS) (% gait cycle), double limb support phase (DLS) (% gait cycle), and foot placement angle (deg).
2.4. Statistical Analysis
GAITRite scores were measured with continuous scales. To determine any gradient of change or systematic differences between the test-retest scores, Wilcoxon rank tests were used. We used intraclass correlation coefficients (ICC) that produce measures of absolute agreement of values within subjects. The model used was a tow-way model for single measure reliability (ICC(3,1)). To quantify the agreement between measurements we estimated repeatability coefficient which is defined by: 1.96*sqr(2)*within subject SD. To estimate the within subject SD we fitted a one way analysis of variance (ANOVA) model. The estimated repeatability coefficient means that the absolute difference between any two measurements made by the same tool are estimated to be no greater than the repeatability coefficient on 95% of occasions. We calculated the 95% C.I. using the Chi-Square distribution. Use of the repeatability coefficient relies on the difference between measurements being approximately normally distributed. Therefore, we used normal plots as well as non-parametric Kolmogorov-Smirnov tests to check for the normality of the differences. We used Spearman correlation tests to estimate the linear relationship of the absolute difference between tests in gait velocity and cadence and BMI and age. Tests were considered significant at a significance level of 5% and a power of at least 80%.

Results
3.1. Patient Characteristics
This study was conducted on 41 patients (13 males and 28 females, 31.7% and 68.3%, respectively) diagnosed with bilateral medial compartment knee OA. Patients' mean ± SD age was 66.7 ± 10.7 years (range 45-88) and BMI was 30.8 ± 4.9 kg/m2 (range 20.8-41.4). In females, mean ± SD age was 66.9 ± 9.7) years (range 45-83) and BMI kg/m2 was 31.7 ± 5.5 (range 20.8-41.4). In males, mean ± SD age was 66.3 ± 13 years (range 44-88) and BMI kg/m2 was 29.6 ± 3 (range 25.5-31.1). The mean BMI of the study population was slightly above average. This was not surprising since patients suffering from knee OA often have an elevated BMI due to inactiveness (13).
3.2. Test-Retest Reliability
Intraclass correlation coefficients (ICC) for all gait parameters were statistically significant and were above 0.85. Both males and females demonstrated high ICC results. Males, however, demonstrated higher results in all gait parameters compared to females (Table 1). Spearman correlations were very similar to ICC (not demonstrated).
All spatio-temporal gait parameters showed similar results in both sessions, except for velocity, normalized velocity and cadence (Table 2). 68% of patients demonstrated a decrease in absolute velocity and normalized velocity. 66% of patients demonstrated a decrease in cadence. The repeatability coefficients demonstrate small expected absolute differences between any two measurements made by the GAITRite system (Table 2).
A calculation of the correlation between the absolute difference between the test-retest scores for velocity and cadence with the age and BMI of the patients was conducted in order to determine the reason for the test-retest difference. A significant moderate correlation was found between the absolute difference in velocity and cadence and the level of BMI of the patients. In other words, patients that demonstrated a decrease in gait velocity and cadence following the resting period, correlated with higher BMI values (Table 3).

Discussion
The aim of this study was to test the reliability of the GAITRite system in patients with knee OA. The results showed that test-retest reliability of the mat in all gait parameters was high. The study results are in agreement with previous studies that have also found high reliability of the GAITRite mat in the velocity parameters and in the gait cycle phases. The correlation of the base of support and of the foot placement angle in previous works, however, is unclear (4, 5). In the present study there was a high correlation of the foot placement angle parameter between the two tests. A possible explanation for the high correlation may be found in the association of changes in foot placement angle with knee OA specifically (7). It is known that patients with medial compartment knee OA have higher adduction moments compared to healthy populations (6). It is also known that a greater toe out angle during gait shifts the ground reaction force vector closer to the center of the knee, thus reducing the adduction moment (14). It may be assumed that knee OA patients consistently adopt the greater toe out angle to avoid high adduction moments and pain, and therefore should have a more consistent foot angle placement compared to a healthy population.
The present study also revealed significantly lower values in the second test in the absolute velocity, normalized velocity, and cadence. These shifts were consistent among the entire group and therefore do not impinge on the reliability of the GAITRite. These differences, however, strengthens the importance of evaluating the validity and reliability of the measuring tool. A possible explanation for these findings is concealed in the characteristics of OA disease. Patient with knee OA suffer from joint stiffness, which appears after a period of inactivity, such as prolonged sitting (15). In the current study, patients who completed the first test were asked to sit and rest for 15 minutes before performing the second test. Patients were then asked to rise and initiate walking without having time to adjust. The new gait strategy adopted by the patients was to reduce cadence frequency, rather than shorten step length, which resulted in lower velocity (Velocity = Cadence x Step Length). The distributions of the severity of OA (KL assessment) in patients who demonstrated an increase in velocity and cadence compared to patients who demonstrated a decrease in these parameters were similar; therefore, the changes in the above parameters are not a result of OA severity Future studies should further examine the effect of the sitting period on gait characteristics of patients with knee OA by incorporating kinematic analyses of gait patterns. While these changes don't affect the reliability of the GAITRite, a clinician must take these effects into consideration when using the GAITRite to evaluate the gait of patients with knee OA. For example, a rested patient demonstrating low velocity and cadence on a gait test may have higher values if they undergo the test without resting. The results of this study also suggest that the aforementioned effects of resting on cadence and velocity are even more likely if the patient has a high BMI. With this considered, it may be beneficial for patients with knee OA to walk before conducting a gait test so that they are accommodated to the test procedure and free of joint stiffness that may influence their gait.
In conclusion, our results demonstrate high reliability of the computerized mat as a tool in assessing the gait spatio-temporal parameters in patients with knee OA.

Acknowledgments
The authors thank Nira Koren-Morag, PhD, for statistical analysis assistance and Mrs. Myrna Perlmutter for editorial assistance.

Fig. 2. Leg Length Measurement. Leg length was measured in a standardized method with the patients standing in an upright position. The clinician located the greater trochanter and lateral malleolus on each leg and used measuring tape to measure the distance from the greater trochanter to the floor passing through the lateral malleolus.

Table 1. Test-retest intraclass correlation coefficient (ICC) and respective confidence intervals (C.I.) for all patients and for genders.

	ICCa	95%C.I.	ICCa	95%C.I.	ICCa	95%C.I.
	All (N=41)		Males (N=13)		Females (N=28)
Velocity Parameters
Velocity (cm/s)	0.89	0.80-0.94	0.98	0.95-0.99	0.82	0.64-0.92
Normalized velocity (cm/s/leg length)	0.86	0.75-0.93	0.98	0.93-0.99	0.82	0.63-0.91
Cadence (steps/min)	0.94	0.67-0.91	0.95	0.86-0.99	0.80	0.56-0.91
Step Length Left (cm)	0.95	0.90-0.97	0.98	0.94-0.99	0.91	0.82-0.96
Step Length Right (cm)	0.95	0.91-0.97	0.98	0.94-0.99	0.92	0.84-0.96
Normalized Step Length Left (cm/leg length)	0.91	0.84-0.95	0.97	0.90-0.99	0.89	0.77-0.95
Normalized Step Length Right (cm/leg length)	0.93	0.87-0.96	0.97	0.91-0.99	0.90	0.80-0.95
Gait Cycle Phase Parameters
Swing Left (% Gait Cycle)	0.91	0.83-0.95	0.95	0.86-0.99	0.86	0.77-0.95
Swing Right (% Gait Cycle)	0.91	0.83-0.95	0.93	0.80-0.98	0.88	0.77-0.95
Stance Left (% Gait Cycle)	0.90	0.82-0.95	0.95	0.85-0.99	0.88	0.75-0.94
Stance Right (% Gait Cycle)	0.88	0.79-0.93	0.93	0.79-0.98	0.85	0.70-0.93
Single Support Left (% Gait Cycle)	0.89	0.80-0.94	0.91	0.75-0.97	0.87	0.73-0.94
Single Support Right (% Gait Cycle)	0.91	0.84-0.95	0.96	0.88-0.99	0.89	0.78-0.95
Double Support Left (% Gait Cycle)	0.90	0.81-0.94	0.94	0.81-0.98	0.87	0.74-0.94
Double Support Right (% Gait Cycle)	0.89	0.80-0.94	0.91	0.75-0.97	0.87	0.73-0.94
Other Parameters
Base of Support Left (cm)	0.91	0.84-0.95	0.91	0.73-0.97	0.91	0.82-0.96
Base of Support Right (cm)	0.92	0.85-0.96	0.91	0.73-0.97	0.92	0.83-0.96
Toe In/Out Left (deg)	0.95	0.91-0.98	0.96	0.87-0.99	0.95	0.85-0.98
Toe In/Out Right (deg)	0.92	0.86-0.96	0.97	0.91-0.99	0.92	0.84-0.96

a Interclass correlation coefficient (ICC) produces measures of consistency or agreement of values within cases.

Table 2. Mean (±SD) values of spatio-temporal gait parameters for the two tests and repeatability coefficient between the two tests.
	Test 1	Test 2	pa	Repeatability Coefficient	95%C.I.
Velocity Parameters
Velocity (cm/s)	90.4 (±24.8)	86.5 (±23)	0.007	7.61	6.31-9.92
Normalized velocity	1.08 (±0.2)	1.03 (±0.2)	0.009	0.09	0.08-0.12
(cm/s/leg length)
Cadence (steps/min)	103.9 (±12.6)	100.8 (±12.1)	0.001	6.03	5.00-7.81
Step Length Left (cm)	51.6 (±9.6)	50.9 (±9.3)	0.12	1.44	1.19-1.86
Step Length Right (cm)	51.2 (±10)	50.7 (±9.7)	0.25	0.97	0.80-1.25
Normalized Step Length Left	0.62 (±0.09)	0.61 (±0.09)	0.14	0.02	0.02-0.03
(cm/leg length)
Normalized Step Length Right	0.61 (±0.1)	0.60 (±0.09)	0.29	0.01	0.01-0.02
(cm/leg length)
Gait Cycle Phase Parameters
Swing Left (% Gait Cycle)	36.1 (±2.8)	36.0 (±2.6)	0.65	0.29	0.24-0.37
Swing Right (% Gait Cycle)	35.6 (±3.1)	35.5 (±2.8)	0.30	0.18	0.15-0.24
Stance Left (% Gait Cycle)	63.9 (±2.8)	64.0 (±2.6)	0.70	0.23	0.19-0.30
Stance Right (% Gait Cycle)	64.8 (±3.2)	64.5 (±2.8)	0.30	0.11	0.09-0.14
Single Support Left	35.6 (±3.2)	35.5 (±2.8)	0.32	0.13	0.11-0.17
(% Gait Cycle)
Single Support Right	36.1 (±2.8)	35.9 (±2.6)	0.70	0.32	0.26-0.41
(% Gait Cycle)
Double Support Left	28.3 (±5.3)	28.5 (±4.9)	0.48	0.46	0.16-0.25
(% Gait Cycle)
Double Support Right	28.7 (±5.5)	28.6 (±4.9)	0.95	0.20	0.16-0.25
(% Gait Cycle)
Other Parameters
Base of Support Left (cm)	10.4 (±3.2)	10.4 (±2.9)	0.97	0.01	0.01-0.02
Base of Support Right (cm)	10.5 (±3.0)	10.4 (±3.0)	0.60	0.20	0.17-0.26
Toe In/Out Left (deg)	6.5 (±5.7)	6.0 (±6.0)	0.07	0.98	0.81-1.26
Toe In/Out Right (deg)	7.5 (±4.9)	7.3 (±4.9)	0.59	0.32	0.25-0.42

a Wilcoxon ranked test results indicates significant statistical differences between the two tests. Significance level was set to P<0.05.

Table 3. The correlation between the absolute difference between tests in gait velocity and cadence and BMI and age.

aP≤0.05. A significant moderate correlation was found between the absolute differences in gait velocity and cadence (the difference between the two tests results) and BMI.