Repeatability of a novel laboratory method for characterizing lateral and anterior stability properties of wheelchair seat cushions
Alexandra M. Delazio, David M. Brienza, Patricia E. Karg
University of Pittsburgh, Department of Rehabilitation Science and Technology
Postural instability is a common problem for wheelchair users that can result in increased risk of falling [2, 3]. Individually configured wheelchairs and seat cushions can increase functional reach  and decrease the risk of pressure injury . The seat cushion, as the base of support for the wheelchair user, affects postural stability by resisting moments when the users’ center of mass is displaced. For example, users’ center of mass shifts as they lean during a reach activity or when users encounter a sloped surface. Research exploring the influence of cushion design and setup on pelvic tilt and measures of postural stability are limited.
Standard test methods exist to characterize tissue integrity management properties of wheelchair seat cushions such as immersion, envelopment, hysteresis, impact damping, recovery, and horizontal stiffness . There are currently no methods that focus specifically on the cushion’s ability to resist medial-lateral or anterior-posterior pelvic rotations caused by shifting center of mass, which can compromise stability. Previous studies have shown that the cushion affects users’ pelvic orientations while leaning,  however the results have not been translated to a standardized laboratory test method intended to provide information to compare stability properties between cushions. The method presented here is intended to fill this gap.
This method is designed to evaluate how a wheelchair cushion resists moments at the pelvis. Moments in the test method are created with an off-center load applied to a standard indenter simulating the buttocks and upper thighs. Resulting indenter tilt angles and surface pressure distributions are measured to characterize the cushion response. The intended use of the method is to differentiate stability performance between cushion models and evaluate the effect of setup configurations and the addition of postural inserts on individual cushions. The method has been refined over the past year during which more than 70 distinct cushions of different sizes, constructs, setups and postural inserts were tested.
Thirty 46 cm wide cushions, varying in materials, constructs and thicknesses were selected out of a larger sample to evaluate the method’s intralaboratory repeatability. Of these 30 test cushions, 17 used air-filled bladders for supporting the user, eight were made from combinations of materials and five were foam cushions. Twenty of the samples were classified as Skin Protection cushions according to the Center for Medicare and Medicaid Services coding. A wooden 39 cm Rigid Cushion Loading Indenter (RCLI) covered with a denim material was used for loading the cushions. Specifications for the 39 cm RCLI can be found in RESNA WC-3: 2018 Annex D . The denim outer layer is an addition to the standard indenter and was added to increase friction between the RCLI and cushion to avoid slipping during the mass shift maneuver.
A test fixture was developed with a sliding platform as the test surface that enabled off-center lateral and anterior loading of the RCLI onto a cushion  (Figure 1A). The sliding platform was configured to move a test cushion and a dead load (consisting of an unconstrained RCLI and a steel plate with a combined weight of 18.3 kg) in the medial-lateral (ML) and anterior-posterior (AP) directions using linear bearing sliding mechanisms below the platform. A live load (33.0 kg), delivered via the cushion loading rig, was constrained to act in a vertical direction and applied to the top of the dead load through a pair of swiveling wheels. When the platform was shifted, the live load was offset laterally or anteriorly. Mechanical stops on the sliding mechanisms were set to selectively position the platform such that the live load could be shifted 75 mm laterally to the left of the neutral position on the ML centerline of the indenter (Figure 1B) or shifted 100 mm forward of the neutral position (Figure 1C).The resulting RCLI rotation angles in AP and ML directions were measured using a digital inclinometer (PRO 360, Mitutoyo, Aurora, IL) positioned on the top of the RCLI. A pressure mat (PX100, XSensor, Calgary AB) was placed between the RCLI and the cushion to collect pressure distribution data.
A test cushion was placed onto the platform of the cushion loading rig such that the back edge of the cushion was aligned with the back edge of the platform. The pressure mat was laid over the cushion and the RCLI was placed on top of the pressure mat and the cushion with the most prominent regions of the lower side of the RCLI (the 2 base points corresponding to the ischial tuberosities of a human pelvis) 125 mm ± 10 mm forward of the back edge of the cushion. The steel plate was centered on top of the RCLI and the live load was first applied 145 mm forward from the back edge of the cushion and centered medial-laterally on the RCLI. RCLI orientation was adjusted until the top surface was horizontal ± 1ᵒ about both ML and AP axes. The starting (neutral) orientation angles were recorded. To perform the lateral center of mass shift, the platform supporting the cushion and dead load was shifted medially 75 mm, shifting the live load laterally. To perform anterior center of mass shift, the platform was shifted posteriorly 100 mm, shifting the live load anteriorly. After 60 seconds, ML and AP orientation angles and pressure distribution were recorded. A total of five lateral shift trials and five anterior shift trials were completed for each sample.
Mean change in RCLI orientation from the neutral position to lateral and anterior shift positions were calculated for each cushion from the five repetitions. Mean average pressure, maximum pressure, and contact area were derived from the pressure distribution recordings. Means and standard deviations were computed across all cushions for each outcome parameter.
Reliability of the method was evaluated using repeatability coefficients (RC) to assess measurement error and intraclass correlation coefficients (ICC) to assess relative reliability. Repeatability coefficients serve as an evaluation of precision of each parameter, investigating the difference between trials across the 30 tested cushions within a 95% confidence interval. As described by Bland and Altman, the RC corresponds to the withinsample standard deviation, sw, multiplied by 1.96 and the √2 . The intraclass correlation coefficients served to evaluate the repeatability of the tilt testing trials. Calculated using a two-way mixed effects model with absolute agreement at a 95% confidence interval, the average measure ICC described the reliability when averaging several repetitions of the test . Statistical parameters were calculated using the first three trials and all five trials to determine an optimal number of trials in future testing.
Method Reliability and Repeatability
Tables 1 and 2 summarize the reliability and repeatability of the methodology shown in this paper. RC values for both lateral and anterior tilt angles were low. RC values for average pressure and contact area were also very small for both lateral and anterior shift tests, given their wide ranges of values. RC values for maximum pressure indicate this measure was less precise. ICC values were high across all test metrics and there was negligible difference between the ICC values calculated for the first three trial versus all five trials.
Case Specific Results
In addition to demonstrating reliability of this novel methodology we explored the test’s ability to differentiate between cushions with different constructs, features, and with and without postural inserts. The following three cases demonstrate the value of the test in demonstrating stability changes when varying these parameters.
A comparison between a segmented foam cushion and an interconnected air cell cushion was performed and summarized in Figure 2. Results showed that while lateral tilt angles were similar for both cushion constructs, maximum pressure values for the segmented foam cushion were much higher than those for the interconnected air cell cushion for both sets of testing. A second comparison was made with an interconnected air cell cushion with a feature that allowed for air flow to be selectively restricted between cells on the left and right sides. The results for conditions with and without airflow restriction are shown in Figure 3. Results showed lower tilt and max pressure values with airflow restriction. A third comparison was made for an interconnected air cell cushion both with and without the addition of a postural insert beneath the cushion. Results are summarized in Figure 4 and showed that the addition of a postural insert slightly decreased the lateral tilt angle.
The RC and ICC values demonstrate the reliability and intralaboratory repeatability of the methodology described in this paper. According to the computed RC values, lateral and anterior shift testing can produce reliable results with a precision of less than 0.75° and showed a strong test-retest consistency of results. High RC values for both lateral and anterior shift test maximum pressure results suggest that there is less precision for measured maximum pressure values. High ICC values across all test metrics indicate that the results are highly repeatable across trials. Negligible difference between the ICC values collected for three trials versus five trials suggests that three trials would be sufficient.
The three case studies demonstrate the ability to use the results obtained from this methodology to differentiate between cushions with different constructs, features, and with and without postural inserts. The comparison of cushion constructs, indicated that stability could be achieved with the interconnected air cell cushion with less adverse effects on pressure distribution compared to the foam cushion. The feature comparison suggested that airflow restriction in air cell cushions has a beneficial effect on stability due to a lower tilt response and better pressure distribution. The comparison with and without a postural insert beneath cushion indicated that postural insert increased stability. Although these cases demonstrate the utility of the method to differentiate cushion performance, these results should not be generalized to all cushions fitting these generic descriptions.
This test method evaluation was limited by only accessing data from one test per cushion recorded in successive trials on a single day in one lab. Future intra- and interlaboratory test result comparisons are needed to improve the reliability analysis and determine reproducibility. Also important to note is that this test evaluates one cushion stability factor, the resistance to a load shift, other aspects, such as how the cushion affects a person’s ability to recover from a shift and/or tilt, are not assessed. Currently, this method provides an option for differentiating stability performance between cushion models and evaluating the effect of different setup configurations and use cases. This study suggests the methodology produces reliable and repeatable measurements of tilt angles and surface pressure distributions that characterize the lateral and anterior stability of wheelchair cushions.
The protocol development and data collection were sponsored by Permobil. This analysis and paper were developed under a grant from the National Institute on Disability, Independent Living, and Rehabilitation Research (Grant 90REGE0001). NIDILRR is within the Administration for Community Living (ACL), Department of Health and Human Services (HHS). The contents do not necessarily represent the policy of Permobil, NIDILRR, ACL, or HHS, and you should not assume endorsement by the Federal Government.
 Brienza, D., et al., A randomized clinical trial on preventing pressure ulcers with wheelchair seat cushions. Journal of the American Geriatrics Society, 2010. 58(12): p. 2308-2314.
 Poojary-Mazzotta, P., WHEELCHAIR RELATED FALL RISK AND FUNCTION IN NURSING HOME RESIDENTS: FACTORS RELATED TO WHEELCHAIR FIT. 2017, University of Pittsburgh.
 Okunribido, O.O.J.A.T., Patient safety during assistant propelled wheelchair transfers: the effect of the seat cushion on risk of falling. 2013. 25(1): p. 1-8.
 RESNA, RESNA American National Standard for Wheelchairs - Volume 3: Wheelchair Seating. 2018, Rehabilitation Engineering & Assistive Technology Society of North America: Arlington, VA.
 Koo, T.K., A.F. Mak, Y.J.A.o.P.M. Lee, and Rehabilitation, Posture effect on seating interface biomechanics: comparison between two seating cushions. 1996. 77(1): p. 40-47.
 Bland, J.M. and D.G.J.S.m.i.m.r. Altman, Measuring agreement in method comparison studies. 1999. 8(2): p. 135-160.
 Koo, T.K. and M.Y.J.J.o.c.m. Li, A guideline of selecting and reporting intraclass correlation coefficients for reliability research. 2016. 15(2): p. 155-163.