Introduction

Assessing an individual for the most appropriate wheelchair and wheelchair set-up is important for effective and safe use of the wheelchair. Appropriate assessment tools can result in a well fit and configured manual wheelchair for individuals with spinal cord injury (SCI), which could help reduce shoulder pain and improve wheelchair skills and function. The development and use of clinical tools to help evaluate wheeled mobility is essential, given that 65 million people around the world require a wheelchair for their mobility.1

Two of the most studied methods of evaluating manual wheeling capacity used by clinicians and therapists in prescribing wheelchairs are as follows: (1) the wheelchair skills test (WST)2 from Dalhousie University, which assesses specific wheelchair skills, and (2) the SmartWheel Clinical Protocol (SCP), which uses an instrumented wheel to measure kinetic and temporal–spatial variables. The WST (4.1) is a reliable measure of capacity of skill (test–retest intraclass correlation coefficient=0.90); however, it is less reliable for evaluating whether a skill is performed safely (test–retest intraclass correlation coefficients=0.25).3, 4 The WST has been used by researchers as an outcome measure of total skill, as well as a clinical tool to evaluate individual skills in various adult4 and paediatric5 populations. Many studies have been published on the use of wheelchair skills assessments to evaluate an individual’s capacity to propel and manoeuvre a manual wheelchair in different environmental conditions.6, 7, 8, 9 The skills included in the WST were modelled on activities of daily living and not surprisingly have been shown to predict quality of life and community integration.10

In addition to wheelchair skills tests that have gained popularity among researchers and clinicians, instrumented measurement wheels have been used by researchers and rehabilitation centres to record and evaluate handrim kinetic and temporal-spatial data during wheeling.11, 12 Commercially available instrumented wheels (for example, SmartWheel11 and Optipush12) have been developed with software to enable clinicians to obtain biomechanical data about how a client propels their wheelchair. The SCP evaluates biomechanical variables while a client wheels on three surfaces (tile, carpet and 5° ramp). A normative clinical database was created from SCP data to describe wheelchair propulsion in adults with SCI.11

The SCP is a reliable tool for wheelchair users with SCI.13 High intersession reliability was determined for force, push frequency, push angle and speed (for a single trial) with intraclass correlation coefficients ranging from 0.70 to 0.85. Intersession reliability increased when an average of five trials were computed. The SCP average of five trials on the 5º ramp surface increased the reliability to a range of 0.97–0.98.13

Askari et al.14 created the wheelchair propulsion test (WPT) in an attempt to be more inclusive of all wheelchair users including hand and foot propellers, and young children whose small diameter wheels may not be compatible with an instrumented wheel. The WPT was created to be a ‘low tech’ measurement tool, which could be used in less-resourced settings as it only requires a tape measure and stop-watch to calculate speed (m s−1), push frequency (#pushes s−1) and push effectiveness (m per push) over 10 m of wheeling. This tool has a high inter-rater reliability (intraclass correlation coefficients=0.96, 0.83 and 0.80 for speed, push frequency and push effectiveness, respectively). However, intersession (that is, test–retest) reliability for the WPT has not yet been determined. The authors compared the WPT to an instrumented wheel and found that speed and push frequency, measured with a tape measure and stop-watch, correlated well with the values obtained from their instrumented wheel. Interestingly, push effectiveness was significantly lower when evaluated with the stop-watch method compared with the values obtained with an instrumented wheel, thus an instrumented wheel may be more accurate.14

Although there are reliable and valid tools for evaluating wheelchair propulsion, it is unclear whether they are related to each other. The primary objective of this study was to examine the relationship between the SCP variables (weight-normalised peak force, speed, push frequency and mechanical effectiveness) and the WST score among a population of adults and children with SCI. The secondary objective was to examine the relationship between the WST score and the WPT’s push effectiveness in this same population. We hypothesised that SCP variables would be predictive of overall skill and specifically that the WST score and push effectiveness would be highly correlated.

Materials and methods

Participants

Individuals who had an SCI (traumatic/non-traumatic) and used a manual wheelchair for >50% of their daily mobility were recruited to participate in this study. Participants were recruited via advertisements at adult and paediatric outpatient rehabilitation centres. Inclusion criteria included the following: 10–65 years of age, able to understand English, follow directions and have a wheelchair with 22′, 24′ or 25′ wheels with quick release axles. We certify that all applicable institutional and governmental regulations concerning ethical use of human volunteers were followed during the course of the study. Demographic and clinical background information were obtained from participants. This information included age, sex, diagnosis/ level of injury, height, weight and type of wheelchair used (Table 1).

Table 1 Participant demographics (n=24)
Full size table

Data collection—WST (4.1)

The WST Version 4.1 is an objective test of wheelchair skills.4 The WST evaluates performance and safety for 32 skills including indoor skills (for example, forward and backward wheeling), community skills (for example, wheeling over potholes) and more advanced skills (for example, skills involving wheelies). Scoring of the WST involves calculation of a total percentage score where the numerator is the total raw score (that is, the number of skills awarded a pass) and the denominator is the number of applicable skills for both capacity and safety. To ensure safety during all skills, an additional person acting as a spotter used a spotter strap attached to the wheelchair’s rear axle to prevent tips. Testing was conducted by one of the two co-investigators trained in implementing the WST. Our analysis of the WST excluded three skills, including (1) gets from ground into wheelchair, (2) ascends stairs and (3) descends stairs. These skills were not performed because of time constraints or lack of adequate spotting for safety.

Data collection—SmartWheel

The SmartWheel (OutFront, Mesa, AZ, USA) was affixed to the participant’s wheelchair on the side of their dominant arm. Participants were provided up to 5 min to familiarise themselves with the weight of the SmartWheel. The instrumented wheel uses six strain gauges to measure three-dimensional forces and moments applied to the handrim. The SCP requires participants to wheel at a self-selected speed for 10 m or 10 s, whichever occurs first, on three surfaces: tile, carpet and a 5° ramp. The SCP software calculated several biomechanical variables including weight-normalised peak force (%body weight), speed (m s−1), push frequency (#pushes s−1) and mechanical effectiveness (tangential force per total force).11 In addition, the WPT's push effectiveness (m per push) was determined using the SmartWheel data for distance travelled and number of pushes applied to the handrim.

Statistical analysis

All statistical analyses were performed with the statistical software R (version 3.0.2, R Foundation for Statistical Computing, Vienna, Austria).15 Multiple regression analyses were used to determine which SCP variables were predictive of the WST score, with separate analyses for each of the three surfaces (tile, carpet and ramp). Pearson’s correlations were performed between push effectiveness and the WST score for each of the three surfaces. Significance was set to P<0.05. Independent sample t-tests were used to compare variables between adult and paediatric participants to ensure that they were homogeneous and could be combined into one group.

Two of the SCP variables, peak force (%body weight) and speed (m s−1), are often highly correlated and not surprisingly demonstrated multicollinearity, with the variation inflation factor values exceeding 2.0. Therefore, we opted to exclude peak force from the variable selection phase of the model building. To develop a parsimonious model, we removed the variable(s) with the highest (nonsignificant) P-value(s) and compared the nested models using analysis of variance. Tolerance level for variable selection was set to α=0.1.

Results

Demographics for the 24 participants including age, sex, diagnosis/level of lesion and wheelchair type are shown in Table 1. Kinetic, temporal-spatial wheeling variables and WST scores for both adult (18–52 years) and paediatric (10–17 years) participants showed no differences between age groups except for peak force (N) and push angle (°), which were significantly larger in the adult group compared to the paediatric group (Table 2). The average number of pushes from the SCP included in the multiple regression analyses was 7.1 (range 5–10), 8.6 (range 5–11) and 9.3 (5–10) for tile, ramp and carpet, respectively.

Table 2 Wheeling parameters from the SmartWheel Clinical Protocol (tile) and wheelchair skills test scores for paediatric and adult groups
Full size table

For the relationship between SCP-TILE and WST, speed and mechanical effectiveness explained 36% of the variance in the WST score (Table 3). For the SCP-RAMP and SCP-CARPET, speed alone explained 58 and 37% of the variance in WST score, respectively (Table 3). Push effectiveness and WST score were strongly (r>0.5) correlated for all surfaces (Table 4).

Table 3 The relationship between SmartWheel Clinical Protocol variables and WST score (df=23)
Full size table
Table 4 Correlations between WST score and push effectiveness (df=23)
Full size table

Discussion

This study is the first to our knowledge that explores the relationships between SCP outcomes and wheelchair skills. The primary purpose of this study was to evaluate the relationship between the SCP variables and the WST score among manual wheelchair users with SCI. Our models showed that regardless of wheeling surface, speed was a significant predictor of WST score. Pradon et al.16 also showed correlations between WST score and self-selected wheeling speed (r=0.57; r2=0.32) and maximal wheeling speed (r=0.72; r2=0.51). Although the ramp and carpet wheeling trials may not require maximal effort, they do require increased effort and power output compared with the tile trials. Wheeling on a ramp or carpet may be closer in intensity to the maximal wheeling speed test on the tile surface. If a clinician needed to minimise time spent on the SCP, using only the ramp or carpet trials could be feasible in a clinical setting and may also be predictive of a client’s wheelchair skills.

The WPT may be a useful tool especially for less-resourced settings; however, the outcome’s sensitivity to change is important to consider. Knowing the minimal detectable change of the WPT, specifically the push effectiveness, will be useful in determining whether it will be adopted as a valuable outcome measure for evaluating the impact of an intervention.

Previous research has shown that wheeling biomechanics is dependent upon wheeling surface (for example, tile, ramp, and carpet).17 Koontz et al.17 found that when wheeling up a ramp, wheelchair users travelled shorter distances per push, used more pushes and had a slower average velocity, which they postulated to be the result of increased deceleration between pushes. Thus, it must be acknowledged that wheeling biomechanics performed on ramp or carpet are different from wheeling on tile.

Our study is one of the first to include adult and paediatric populations in the same cohort. Research in this field is often plagued with small sample sizes, and this study is no exception. However, we believe that the combined adult and paediatric data (n=24) is sufficient to show relationships between the SCP and WST for individuals with SCI who are over 8 years of age. Surprisingly, this is the largest sample of data that evaluates both adults and children using the same protocol. There is not an abundance of literature on wheelchair propulsion in children. One study from the mid 1990's evaluated propulsion in 10 adults and 10 children,18 whereas two other studies in 2013 only assessed children (n=13),19 (n=2).20

We showed surprising similarities in wheeling parameters and skills between adult and paediatric wheelchair users with SCI (Table 2). Only peak force (N) and push angle (º) were different between the two populations with adults using significantly greater peak force (N) and longer push angles. Adults appear to use greater peak force on the basis of their larger mass considering that there were no differences in weight-normalised peak force. Our findings support the work by Bednarczyk et al.18 who concluded that kinematics and kinetics were similar in adult and paediatric populations with SCI. Our paediatric SCI population was fortunate to have ultra lightweight wheelchairs, which were generally well fit. In addition, many children had participated in wheelchair sports, which allowed them to learn higher level skills. The two youngest participants (10 and 11 years old) had the smallest push angles (39.2º and 58º, respectively); however, their wheeling speeds (1.0 and 1.2 m s−1, respectively) were not different compared with the adult speeds. Despite the age differences between the two populations evaluated in this study, the data support combining these two groups for the purpose of our analyses (Table 2). Although we standardized tyre pressure by pumping participants’ tyres to the maximum recommended pressure, there may be confounding variables that we did not control for, such as wheelchair training experience, self-efficacy, level of SCI and wheelchair type or maintenance.21 Four participants (one paediatric) received 100% on the WST, which may have introduced a ceiling effect into our results; however, there were no violations of any of the model assumptions that would bring into question the validity of our results. The ceiling effect was partially due to our exclusion of two very difficult skills (ascending and descending stairs). Future studies should evaluate all skills from the WST in order to reduce this ceiling effect and incorporate the newest version of the WST (4.2).2

In conclusion, there appear to be predictive relationships between wheeling parameters and wheelchair skills for individuals with SCI. Speed on tile and carpet and speed and mechanical effectiveness on ramp were significantly predictive of the WST score. Push effectiveness was also highly correlated (r>0.5) with WST score and may be a valuable outcome if skills testing is not available. Depending on the time or financial resources available to a clinician, performing a speed test on a ramp or carpet may be a valid and reliable method to follow a patient's wheelchair propulsion progress. However, the authors believe that the evaluation of biomechanical variables cannot replace a comprehensive evaluation of skills and that each tool has value in evaluating wheelchair propulsion.

DATA ARCHIVING

There were no data to deposit.