The reliability of kinetic and kinematic variables used to analyse normal running gait

Gait and Posture 14 (2001) 98 – 103 www.elsevier.com/locate/gaitpost

The reliability of kinetic and kinematic variables used to analyse normal running gait Ceri E. Diss * Uni6ersity of Surrey Roehampton School of Sport, Exercise and Leisure, West Hill, London, SW 15 3SN, UK Received 14 November 2000; received in revised form 3 March 2001; accepted 2 April 2001

Abstract The purpose of this study was to assess the reliability of 24 kinetic and kinematic variables from three synchronized systems used to represent normal running gait. Five male runners (mean 23.4 years, mass 80.2 kg) ran down a runway at a constant velocity (3.5–4.0 m/s). This was repeated until 10 acceptable trials had been performed which was then repeated 7 days later. The mean of the 10 trials was used for kinetic analysis and the mean of 5 trials for the kinematics. All of the kinematic variables achieved a reliability greater than 0.93. The 6 variables that were able to demonstrate a high reliability ( \ 0.94) from a single trial came from all systems. This suggests that variables and systems used to collect and analyse gait should be assessed for their reliability using the population to be studied before actual data collection. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Gait; Kinematics; Kinetics; Levels of confidence; Reliability

1. Introduction Many studies have considered the mechanics of gait patterns to produce ‘norm’ data, reduce and prevent injury, evaluate the function of footwear and to improve performance [1 – 8]. Analysing kinetic and kinematic variables associated with gait gives an insight into the cause of movement that might be either detrimental to or produce an improvement in performance. Researchers have drawn conclusions on an individual’s gait from data taken over a number of trials for each given experimental condition. There has been much debate as to the number of trials a subject should perform to reduce the variability that inevitably exists between trials and data collection sessions. Bates et al. [9] found that when evaluating small kinematic and temporal changes during treadmill running, average values taken from a number of trials were more valuable than a single trial for an individual’s gait. Previous studies have shown that at least three gait cycles should be examined to obtain reliable measures [10 – 12]. * Corresponding author. Tel.: + 44-20-83923541; fax: +44-2083923749.

Bates et al. [13] used ten successful trials for each subject/condition to develop a model of selective ground reaction forces that could be used to represent an individual’s gait. Bates et al. [14] also stated that a minimum of eight trials was required to establish normal patterns under experimental conditions. DeVita et al. [15] found that to obtain stable mean parameters at least 25 trials were required, especially when analysing footwear using ground reaction force data. Other investigators have used fewer trials [6–8]. Van Woensel et al. [16] found that ‘time taken to maximum pronation’, was an unreliable variable when studying rear foot motion even when an average of 10 trials was used. Winter [12] concluded that subjects demonstrated low variability for kinematic and ground reaction force data (7–20%) but high variability for moment force patterns at the hip and the knee (67 – 72%). Lees et al. [17] found that inexperienced runners showed greater variability in braking and propulsive impulses compared with experienced runners. This conclusion was based on data from three data collection sessions where 20 trials/session were used for analysis. Bates et al. [13] also stated that runners were mechanically and anatomically different and therefore one would expect intra/inter subject variability.

0966-6362/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 9 6 6 - 6 3 6 2 ( 0 1 ) 0 0 1 2 5 - 4

C.E. Diss / Gait and Posture 14 (2001) 98–103

In some cases, certain variables and subjects will show inconsistency between trials even under exactly the same conditions, regardless of the number of trials recorded. DeVita et al. [15] stated that researchers needed to recognise that variability under test conditions could result from ‘normal performance variability’. They stated that ‘multiple trials will be necessary for most movements, however the exact number may vary with activity.’ Therefore, conclusions drawn from a limited number of trials need to be treated with caution especially when variables are seen to change from small alterations to the independent variables. The changes seen in the criterion variable may be a result of its (or the subject’s) variability and not a consequence of an intervention. Therefore, the variable nature of the subject and variables being used for analysis should be assessed. Ideally, this would take place with each subject and experimental condition but in reality this would be time consuming and costly. However, assessing the variable nature of the variables associated with certain movement patterns is more practicable particularly when time is limited or if the subject is unable to perform multiple trials. The aim of this study was to examine the reliability of 24 kinematic and kinetic variables from three synchronized systems that are commonly used to analyse an individual’s normal running gait and to ascertain the number of trials required for each variable to obtain accurate data.

2. Methods Five male runners (mean age=23.4 years, mean mass =80.2 kg) were tested on two occasions, 7 days apart. There was no intervention that would have caused an alteration in their gait pattern between the data collection sessions. All subjects were rear foot strikers and during the data collection each subject wore his normal running shoes. All shoes were qualitatively assessed for rigidity of the heel counter and compliance of the cushioning element of the rear section. Informed consent was obtained before data collection. The procedure for each data collection day was identical. Each subject jogged for 8 min on a Woodway (model PPS 55med) before undergoing his normal stretching routine. A familiarisation period then followed so that a constant running velocity was maintained down a 20 m runway, ensuring a right foot strike on the force platform. This time varied between subjects as it was important that the subject did not alter his gait pattern in any way to strike the plate and that he felt comfortable with the approach. After familiarization, six reflective markers were placed on the subject’s right and left hip, knee and ankle joint centres and a further two were placed on the

99

hallux (right and left). four markers were placed on the right rear lower leg in accordance with the protocol reported by Nigg [2]. The subject was permitted additional familarisation, if required. Three synchronized systems were used for data collection (Diagram 1). A Kistler force plate (type 9281B11) operating at 500 Hz was flush to the floor approximately 15 m along the runway. Force data were captured and analysed via an amplifier (type 9865), analogue-to-digital converter (C10 DA516F) and Kistler’s Bioware software (version 2.0). Two genlocked Panasonic F15 video cameras operating at 50 field/s, shutter speed 1/500 s, were mounted 1 m high and placed at 30° either side of the plate at a distance of 6.5 m. A Vicon 12 point calibration frame (360 mm× 400 mm× 498.35 mm) was placed on the force plate before data collection that ensure a large image. An Acorn Archimedes 440 computer using a moving i-mage (London, UK) digitisation board (resolution 1024×512) and Kine Analysis software (Manchester, UK) was used to create 3-dimensional coordinates of the camera views by Direct Linear Transformation algorithm [18]. A four-segment model was created to yield the calcaneal rear foot angle by digitising the markers placed on the rear of the lower leg. Generalised cross-validated quintic spline [19] was applied to the file for smoothing. A Vicon 140 four camera automated system operating at 50 Hz was used to collect joint angle data from the eight reflective markers. The infrared cameras were clamped in the corners of a static grid placed above the force platform. Again, using the Vicon calibration frame, the field of view was calibrated. Joint angles were analysed after data collection and reconstruction (Fig. 1). Five acceptable trials were obtained during each data collection session and a further five trials for force data. The acceptability was based upon a ‘natural’ foot strike and a running velocity of 3.5–4.0 m s − 1. The velocity was measure by a Griffin timing unit using four cells placed either side of the plate 1.5 m apart. The mean and standard deviation for all subjects from both data collection days were calculated for all 24 variables. Reliability (R(10) or R(5)) of the mean of ten/five trials was calculated using the ANOVA with: R= (MSb − MSw )/MSb where MSb is the mean square between subjects and MSw is the mean square value within subjects plus error measurement. The reliability of one trial R(1) can be determined from the following equation: R(1)= (MSb − MSw )/(MSb + ((Nt − 1)MSw )) where Nt is the number of repeated measures (10 or 5). To calculate the number of trials required to achieve a certain level of reliability the Spearman-Brown prophecy formula can be used:

100

C.E. Diss / Gait and Posture 14 (2001) 98–103

Fig. 1. Experimental setup.

N =(rc (1− R))/(R(1 −rc )) where R is the calculated reliability from the number of trials (R(1)), rc is the level of reliability required and N is the number of trials required in order to achieve such a reliability.

3. Results The mean and standard deviation for the five subjects, from both data collection sessions is shown in Table 1. The variables chosen for analysis are a selection taken from a large number that can be used to analyse an individuals’ gait [1]. Stance time showed that each subject remained in contact with the ground for almost the identical amount of time for each trial. The angle of inversion showed that the calcaneus was tilted at a mean value of 5.96°. This gave an indication of the position of the foot upon impact. Mean eversion angle was − 9.07° which gave the position of the calcaneus associated with supination. The time between these positions and the total range of calcaneal motion represented the speed of this joint’s motion. Lower body right side kinematics showed the body’s position at key absorption and toe-off phases during stance. Total range of motion values was used to ascertain the angles through which the joint moved. Table 2 shows the reliability of the ground reaction force (GRF) variables when 10 trials and a single trial were taken. When the mean of 10 trials was taken, the lowest reliability value was 0.93, which occurred for decay rate from the second data collection session. The highest reliability was from 10 trials and a single data trial for stance time (1.00). Zero fore-aft shear and decay rate had the lowest reliability when one single trial was taken (0.59 and 0.57, respectively).

Table 3 shows the reliability of the kinematic data taken from five trials and a single trial. Using the mean of five trials, the lowest reliability (0.93) occurred for maximum eversion angle and ankle joint angle at foot strike from the first data collection session. A reliability of 1.00 occurred from both data collection sessions for time to maximum eversion, ankle joint angle at toe-off and total range of ankle motion for both. Time to maximum eversion demonstrated the highest reliability from one single trial (1.00). Maximum eversion angle Table 1 Group mean and standard deviation values for all 24 variables from both data collection sessions (to 2 d.p.) Variable

Mean

sd

Stance time (s) Average vertical force (BW) Loading rate (BW/s) Impact maximum (BW) Thrust maximum (BW) Decay rate (BW/s) Maximum braking (BW) Maximum propulsion (BW) Zero fore-aft shear (%) Maximum medial (BW) Maximum lateral (BW) Inversion angle @ footstrike (°) Maximum eversion angle (°) Time to maximum eversion (s) Total range of calcaneal motion (°) Hip joint angle @ footstrike (°) Hip joint angle @ toe-off (°) Total range of hip motion (°) Knee joint angle @ footstrike (°) Knee joint angle @ toe-off (°) Total range of knee motion (°) Ankle joint angle @ footstrike (°) Ankle joint angle @ toe-off (°) Total range of ankle motion (°)

0.23 1.34 54.53 1.60 2.44 23.12 −0.38 0.35 45.52 0.15 −0.07 5.96 −9.07 0.12 12.07 24.21 9.61 34.11 166.45 170.01 39.26 4.89 21.55 29.46

0.07 0.43 20.18 0.47 0.78 7.72 0.17 0.11 14.34 0.08 0.08 1.92 2.19 0.01 3.30 0.70 0.46 1.67 4.99 7.65 1.27 0.19 0.82 0.10

C.E. Diss / Gait and Posture 14 (2001) 98–103 Table 2 Measured reliability of the mean of 10 trials (R(10)) and calculated reliability for 1 trial (R(1)) from 2 data collection sessions for force data

Stance time (s) Average vertical force (BW) Loading rate (BW/s) Impact maximum (BW) Thrust maximum (BW) Decay rate (BW/s) Maximum braking (BW) Maximum propulsion (BW) Zero fore-aft shear (%) Maximum medial (BW) Maximum lateral (BW)

Data collection 1

Data collection 2

R(10)

R(1)

R(10)

R(1)

1.00 1.00

1.00 0.98

1.00 1.00

1.00 0.96

0.99 0.98 0.97 0.97 0.99 1.00

0.87 0.82 0.76 0.77 0.88 1.00

0.97 0.98 0.99 0.93 0.97 0.98

0.80 0.85 0.89 0.57 0.75 0.85

0.94 0.99 0.98

0.59 0.89 0.82

0.95 0.99 0.98

0.64 0.89 0.85

Table 3 Measured reliability of the mean of 5 trials (R(5)) and calculated reliability for 1 trial (R(1)) from 2 data collection sessions for kinematic data

Inversion angle at footstrike (°) Maximum eversion angle (°) Time to maximum eversion (s) Total range of calcaneal motion (°) Hip joint angle @ footstrike (°) Hip joint angle @ toe-off (°) Total range of hip motion (°) Knee joint angle @ footstrike (°) Knee joint angle @ toe-off (°) Total range of knee motion (°) Ankle joint angle @ footstrike (°) Ankle joint angle @ toe-off (°) Total range of ankle motion (°)

Data collection 1

Data collection 2

R(5)

R(1)

R(5)

R(1)

0.94

0.77

0.94

0.76

0.93

0.72

0.97

0.86

1.00

1.00

1.00

1.00

0.96

0.83

0.99

0.95

0.97

0.86

0.97

0.87

0.99

0.97

0.99

0.94

0.96

0.84

0.96

0.83

0.95

0.78

0.95

0.78

0.96

0.81

0.96

0.84

0.97

0.86

0.95

0.79

0.93

0.72

0.94

0.75

1.00

0.98

1.00

0.98

1.00

0.99

1.00

0.99

and ankle joint at foot strike had reliability values of 0.72 from one single trial taken on the first data collection.

101

Table 4 The number of trials required achieving a certain level of reliability for kinetic and kinematic data for both data collection sessions

Stance time (s) Average vertical force (BW) Loading rate (BW/s) Impact maximum (BW) Thrust maximum (BW) Decay rate (BW/s) Maximum braking (BW) Maximum propulsion (BW) Zero fore-aft shear (%) Maximum medial (BW) Maximum lateral (BW) Inversion angle at footstrike (°) Maximum eversion angle (°) Time to maximum eversion (s) Total range of calcaneal motion (°) Hip joint angle @ footstrike (°) Hip joint angle @ toe-off (°) Total range of hip motion (°) Knee joint angle @ footstrike (°) Knee joint angle @ toe-off (°) Total range of knee motion (°) Ankle joint angle @ footstrike (°) Ankle joint angle @ toe-off (°) Total range of ankle motion (°)

Data collection 1

Data collection 2

R\0.9

R\0.8

R\0.9

R\0.8

1 1

1 1

1 1

1 1

2 2

1 1

3 2

2 1

3

2

2

1

3 2

2 1

7 3

3 2

1

1

2

1

7

3

6

3

2

1

2

1

2

1

2

1

3

2

3

2

4

2

2

1

1

1

1

1

2

1

1

1

2

1

2

1

1

1

1

1

2

1

2

1

3

2

3

2

3

1

2

1

2

1

3

2

4

2

3

2

1

1

1

1

1

1

1

1

Table 4 shows the number of trials required to achieve a reliability greater than 0.8. Decay rate and zero fore-aft shear would require at least 7 trials to achieve a reliability greater than 0.9. For all the other variables the number of trials required to achieve such reliability ranged between 1 and 4.

102

C.E. Diss / Gait and Posture 14 (2001) 98–103

4. Discussion Twenty-four kinematic and kinetic variables were collected to analyse the gait of 5 runners. The stance time was consistent for all subjects. This was to be expected since the velocity of the gait was controlled and trials of between 3.5 m/s and 4.0 m/s were accepted. However, there was not a significant relationship between velocity of gait and stance time as the subjects attempted to produce a consistent stride pattern. Loading rate is an indication of the time taken for the vertical force to rise by the subject’s body weight (BW) and in this study was lower than that previously reported [1]. The first 50 N were ignored because of the positional changes of the foot at initial impact [1]. This was a result of the subjects generating either a smaller force at initial strike or an increased time over which the forces were generated. This caused the gradient of the initial slope on the vertical force-time trace to be less than that reported previously. This could be a result of the modern cushioning properties of footwear, since all contain some kind of absorption material. The mean impact maximum value, which is the first peak seen on the vertical force-time trace, was also lower than previously reported, which could support this view. For a running speed of 3.5– 4.0 m/s, the value is usually between 1.75– 2 BW [1]; however the subjects in this study had a mean value of 1.59 BW. Maximum thrust is normally the second peak seen on a vertical force-time trace. This value increases with speed and is an indication of the subject’s ability to initiate knee extension and initiation of the centre of mass’s (CM) upward movement. Decay rate represents the gradient of the slope from maximum thrust to zero. This was higher than reported in the literature [1], which indicated a steeper slope up to toe-off and that the subjects had the ability to generate a large push-off, which was rapidly transferred to the flight phase of running. Zero fore-aft shear force shows the percentage of stance time spent generating negative horizontal forces. A mean value of 45.52% in the subjects showed that a greater part of stance was spent generating positive horizontal forces. An understanding of the body’s CM position relative to the point of force application can be obtained from an appreciation of the maximum braking and propulsive forces. On impact with the ground the CM was behind the point of force application that caused the subjects to slow down due to the creation of a negative horizontal force. After 45% of stance time the subjects generated positive forces during which the CM moved forward of the point of force application. The negative horizontal impulse equated to the positive since a constant velocity was maintained. As expected, the medial/ lateral forces were relatively small since the motion was predominately in the sagittal plane.

Reliability values above 0.9 are considered to possess a high level of confidence [20]. For this study on five normal runners, the kinetic data had reliability greater than 0.93 when the mean of 10 trials was used. This showed that if a subject from the same population performed 10 trials an accurate representation of their kinetics could be gained. In many circumstance 10 trials is impracticable. However, if a single trial were used for analysis, the reliability falls dramatically for certain variables. From the two separate data collection sessions performed in this study it was found that thrust maximum, decay rate, maximum braking and zero foreaft shear had very low reliability values (0.59– 0.77) if a single trial was used. Decay rate is a function of maximum thrust and time therefore the deviations seen in maximum thrust will affect the decay rate. The reason that thrust maximum may be susceptible to variation could be due to the laboratory setting. The subject’s velocity down the runway was monitored but once they had struck the force platform they knew that the data had been collected and may subconsciously have begun to slow down or ‘push off’ in a different way. This supports the work of Lees & Bouracher (1994) [17] in that the subject should, as far as possible, be unaware of the position of the force platform. A plate placed in the centre of the runway is the optimum position but would require the runway to be over 20 m long if a high constant velocity was required. One would expect a zero fore-aft shear force to demonstrate a low level of reliability from a single trial, because it is dependent upon the position of the CM with respect to the point of force application. Since all the body segments affect the position of CM, a small alteration in one body part will change its position. Kinematic data taken from the mean of 5 trials for 5 normal runners was reliable (R \ 0.93). The high level was maintained for time to maximum eversion, ankle and hip joint angle at toe-off and total range of ankle motion for both data collection days for a single trial (R\ 0.9). However, this reduced to 0.72 for maximum eversion and ankle joint angles at foot strike for a single trial. This may be attributable to the sample rate, which made it difficult to identify the exact time that foot strike or toe-off occurred which in turn would affect the calculated angle. Other sources of error associated with video analysis such as marker placement and digitising are important considerations. This demonstrates that variables assigned for analysis should be carefully chosen if only a single trial is to be recorded. In conclusion, an 80% confidence level can be achieved if the mean of 3 trials is used for analysis when collecting 24 kinematic and kinetic variables from a population of 5 normal runners. To increase the level to 90% at least 7 trials would be required. In both cases decay rate and zero fore-aft shear force require the

C.E. Diss / Gait and Posture 14 (2001) 98–103

greatest number of trials to achieve such a level of confidence. Consideration should be given to the choice of variables, the number of trials required and equipment available to achieve an acceptable level of reliability before data collection.

[11]

[12]

References [1] Cavanagh PR, editor. Biomechanics of Distance Running. Champaign, Ill: Human Kinetic Publishers, 1990. [2] Nigg BM, editor. Biomechanics of Running Shoes. Champaign, Ill: Human Kinetics Publishers, 1986. [3] Nigg BM, Cole GK, Bruggeman GP. Impact forces during heel-toe running. J Appl Biomech 1995;1:407 – 32. [4] Stacoff A, Denoth J, Kaelin X, Stuessi E. Running injuries and shoe construction: some possible relationships. Int J Sport Biomech 1988;4:342 –57. [5] De Wit B, de Clerq D, Lenoir M. The effect of varying midsole hardness on impact forces and foot motion during foot contact in running. J Appl Biomech 1995;11:395 –406. [6] Nigg BM, Bahlsen HA. Influence of heel flare and midsole construction on pronation, supination and impact forces for heel-toe running. Int J Sport Biomech 1988;4:205 –19. [7] Stacoff A, Kaelin X, Stuessi E, Segesser B. The torsion of the foot in running running. Int J Sport Biomech 1989;5:375 – 89. [8] Williams KR, Ziff JL. Changes in distance running mechanics due to systematic variations in running style. Int J Sport Biomech 1991;7:76 –90. [9] Bate BT, Osternig LR, Mason BR, James SL. Functional variability of the lower extremity during the support phase of running. Med Sci Sports Ex 1979;11:328 – 31. [10] Grainger J, Norman C, Winter D, Bobet J. Day to day reproducibility of selected biomechanical variables calculated from

[13]

[14]

[15] [16]

[17]

[18]

[19]

[20]

103

film data. In: Matsui H, Kobayashi K, editors. Biomechanics VIII A & B. Champaign, Ill: Human Kinetic Publishers, 1983:1239 – 47. Hughes J, Pratt L, Linge K, Clarke P, Klenerman L. Reliability of pressure measurements: the EMED F system. Clin Biomech 1991;6:14 – 8. Winter DA. Kinematic & kinetic patterns in human gait: Variability and compensating effects. Hum Mov Sci 1984;3:51 –76. Bates BT, Osternig LR, Sawhill JA, James SL. An assessment of subject variability, subject-shoe interaction, and the evaluation of running shoes using ground reaction forces. J Biomech 1983;16:181 – 91. Bates BT, Osternig LR, Sawhill JA, James SL. Identification of critical Variables Describing Ground Reaction Forces During Running. In: Matsui H, Kobayashi K, editors. Biomechanics VIII A & B. Champaign, Ill: Human Kinetic Publishers, 1983:634 – 40. DeVita P, Bates BT. Intraday reliability of ground reaction force data. Hum Mov Sci 1988;7:73 – 85. Van Woensel, Cavanagh PR. A perturbation study of lower extremity motion during running. Int J Sport Biomech 1992;8:30 – 47. Lees A, Bouracher J. The longitudinal variability of ground reaction forces in experienced and inexperienced runners. Ergonomics 1994;37:197 – 206. Abdel-Aziz YI, Karara HM. Direct linear transformation from comparator coordinates into object space coordinates in closerange photogrammetry. In: Proceedings of the Symposium on Close-Range Photogrammetry. Falls Church, VA: American Society of Photogrammetry, 1971:1 – 18. Woltring HJ. A Fortran package for generalized cross-validatory spline smoothing and differentiation. Adv Eng Softw 1986;8:104 – 13. Vincent WJ. Statistics in Kinesiology. Champaign, Ill: Human Kinetics, 1995.