The French-Canadian WheelCon for manual wheelchair users: paediatric adaptation and preliminary evaluation of its measurement properties

Abstract Purpose To adapt the Wheelchair Use Confidence Scale for Manual Wheelchair Users, French-Canadian Version (WheelCon-M-F) into a paediatric version (WheelCon-M-F-P) and to evaluate its validity. Methods A three-phase process was conducted: (1) item adaptation using secondary analysis of focus group data; (2) item refinement using a think-aloud process; and (3) preliminary validation of the WheelCon-M-F-P (i.e. internal consistency, test-retest reliability, standard error of measurement, smallest real difference, ceiling and floor effects, limits of agreement, and relations with other variables). Results Phase 1: The sample consisted of occupational therapists (n = 9), paediatric manual wheelchair users (PMWUs) (n = 12), and parents of PMWUs (n = 2). Of the 65 WheelCon-M-F items, 35 were removed, 25 modified and 6 were added for the WheelCon-M-F-P. Phase 2: 4 PMWUs helped refine 14 and remove 3 items. Phase 3: 22 PMWUs participated. Cronbach’s alpha, test-retest reliability, standard error of measurement, and smallest real difference were 0.846, 0.818, 3.05, and 8.45 respectively. No ceiling or floor effects were demonstrated. Pearson correlations between the WheelCon-M-F-P and the Wheelchair Skills Test Questionnaire (capacity, confidence, and performance), and the Child Occupational Self-Assessment were 0.688, 0.711, 0.584, and 0.687 respectively. Conclusions This study provides preliminary evidence of a valid and reliable WheelCon-M-F-P. IMPLICATIONS FOR REHABILITATION The Wheelchair Use Confidence Scale for Manual Wheelchair Users, French-Canadian Version (WheelCon-M-F-P) is an outcome measure that can be used clinically with pediatric manual wheelchair users. The WheelCon-M-F-P can help identify modifiable factors associated with wheelchair confidence. Identifying modifiable factors associated with wheelchair confidence can help guide clinicians in establishing a targeted intervention for their pediatric clients. Establishing a targeted intervention can help improve self-efficacy for wheelchair use and social participation of pediatric wheelchair users.


Introduction
Wheelchair confidence, defined as the belief individuals have in their ability to use their wheelchair in a variety of challenging situations, is a key element in the wheelchair provision process [1,2], in wheelchair skills training [3][4][5], and in the abandonment of assistive technology [6].In fact, there exists a high prevalence of adult manual wheelchair users who demonstrate low wheelchair confidence [7], which, in turn, relates to a low level of participation in activities of daily living [7][8][9][10].As well, there often exists a discrepancy between the user's wheelchair confidence and their actual wheelchair skills [9], which can negatively impact the user's daily life.For instance, a user exhibiting a very high level of wheelchair confidence with limited wheelchair skills may be willing to take higher risks in difficult situations despite not having sufficient skills to manoeuvre safely.On the other hand, low wheelchair confidence may impede daily manual wheelchair use despite advanced wheelchair skill, as wheelchair users may refrain from doing activities that they are able to do.Furthermore, higher wheelchair confidence leads to improved social participation, which emphasizes the construct's importance.Fortunately, wheelchair confidence in the adult population has been shown to be a modifiable barrier by wheelchair skills training [10] and represents a useful aspect to target in an intervention plan to improve participation, such as in the peer-led Wheelchair Self-Efficacy Enhanced for Use (WheelSeeU) program [3,11,12].
To evaluate wheelchair confidence, the Wheelchair Use Confidence Scale, abbreviated as WheelCon, was developed as a self-report, subjective outcome measure.This measure covers 6 areas: negotiating the physical environment, performing activities in the manual wheelchair, knowledge and problem solving, advocacy, managing social situations, and managing emotions.To date, there exists a WheelCon for adult manual wheelchair users (WheelCon-M) [8] and adult power wheelchair users (WheelCon-P) [13], a short form for each of these versions (WheelCon-M-SF [14] and WheelCon-P-SF [15]), an Italian version of the WheelCon-M-SF [16], and a French-Canadian version of the WheelCon-M (WheelCon-M-F) [17].In addition, the WheelCon-M was adapted into a Dutch youth version, the WheelCon-M-D-Y.This version has demonstrated preliminary evidence of internal consistency and construct validity, but it includes items related only to the "negotiating the physical environment" area of wheelchair confidence [18].
A recent qualitative study highlighted the multifaceted aspects of paediatric manual wheelchair confidence according to the perceptions of occupational therapists, parents, and paediatric manual wheelchair users (PMWUs) [19].The findings suggested that paediatric manual wheelchair confidence goes beyond solely negotiating obstacles and includes managing social situations and emotions.These findings, combined with our knowledge of the importance of all six areas of wheelchair confidence among adult manual wheelchair users, support the need for a paediatric version of the WheelCon.Thus, the objectives of this study were to: (1) adapt the WheelCon-M-F into a paediatric version (WheelCon-M-F-P) and ( 2) evaluate the paediatric version of the WheelCon-M-F validity evidence based on response processes, the internal structure (i.e.internal consistency, test-retest reliability, responsiveness (i.e. standard error of measurement, smallest real difference), ceiling and floor effects and limits of agreement), and relations with other variables.We hypothesized that the WheelCon-M-F-P would: (i) be internally consistent (Cronbach's alpha of ≥ 0.80); (ii) have test-retest reliability (intraclass correlation coefficient [ICC 1.1 ] ≥ 0.80); (iii) have a positive moderate correlation with the scores of both Wheelchair Skills Test-Questionnaire (WST-Q) and the Child Occupational Self-Assessment (COSA); and (iv) have a positive weak correlation with participants' age and years of wheelchair experience.

Methods
This study involved three sequential phases, each with its own objective (Figure 1).Phase 1 involved the adaptation of the WheelCon-M-F into a paediatric version using secondary analyses [20] of existing data.In Phase 2, we conducted an item refinement of the WheelCon-M-F-P using a Think Aloud approach [21].In Phase 3, we evaluated the WheelCon-M-F-P validity evidence using a test-retest study design [22].The study was approved by the Sainte-Justine University Hospital Research Centre Ethics Board.All participants provided informed consent or assent (for the paediatric participants 18 years of age or younger).

Design
A secondary analysis of existing focus group and qualitative interview data was conducted to adapt the WheelCon-M-F adult version into the paediatric version.

Procedure and data analyses
The data were from focus groups conducted with occupational therapists and PMWUs, and qualitative interviews conducted with parents of PMWUs, all of which had the aim of exploring perspectives on wheelchair confidence among PMWUs [19].The secondary data analysis plan involved a deductive qualitative approach [23] whereby the data were analysed to identify content related to each WheelCon-M-F item with respect to its relevance for measuring wheelchair confidence among children.If the item was deemed relevant, its need for adaptation to either simplify the item to enhance its understandability for children or to enhance its paediatric specificity was determined.The data was also analysed to identify additional content (i.e.items) to be included in the paediatric version.This analysis was conducted by one study investigator (a Master's in occupational therapy student) and confirmed by a second investigator (an occupational therapist and assistant professor with expertise in wheelchair confidence).In addition, the response scale was adapted to be used by the paediatric population based on data collected from a previous study [19].

Design
A think-aloud process [21] was used to assess the response processes for each WheelCon-M-F-P item.This method allowed us to determine if each item was interpreted as intended and to resolve any identified issues prior to the evaluation of its measurement properties.

Participants
To recruit potential participants, a short presentation explaining the study was given to the occupational therapists at the Centre Intégré Universitaire de Santé et de Services Sociaux de la Capitale-Nationale (CIUSSS-CN; Quebec, Quebec) and at the Marie-Enfant Rehabilitation Center (CRME; Montreal, Quebec) and two affiliated schools, École Victor-Doré (primary school) and École Joseph-Charbonneau (high school).Following the presentation, follow-up emails were sent that provided the study's letter of information for the therapists to provide to their clients (or their client's parents) who fit the study criteria.To be included, children needed to: (1) be between the ages of 5 and 18; (2) use a manual wheelchair at least 4 hours per day for a minimum of 6 months prior to the study; (3) live in Montréal or Québec City; and (4) be able to read, write, and communicate verbally in French.

Procedure
Participants began the data collection session by completing a socio-demographic questionnaire, which collected information including age, sex, diagnosis, years of wheelchair experience, primary language, and primary location of wheelchair use.Next, the WheelCon-M-F-P was provided to the participant, accompanied by a description of the measure, as well as administration and response scale instructions.The think-aloud process (i.e. to say whatever comes into their mind) was then described and demonstrated to the participant.The study investigator remained silent throughout the completion of the measure to allow the uninterrupted flow of thoughts.The verbal cue, "continue to think aloud," was given only when participants stopped verbalizing their thoughts.All participants' responses were audio recorded and transcribed verbatim.

Data analyses
Demographics were analysed using descriptive statistics.Two study investigators coded the "think aloud" responses independently using Fowler and Cannell's behavioural code categories [24], including: "adequate" (i.e.participant gives the answer that meets item objective); "clarification" (i.e.participant asks for clarification of item, or makes statement indicating uncertainty about item meaning); "qualified" (i.e.participant gives an answer that meets item objective, but is qualified to indicate uncertainty about accuracy); "inadequate" (i.e.participant gives an answer that does not meet item objective); and "refusal to answer" (i.e.participant refuses to answer the question).Discrepancies in coding were resolved through discussion until a consensus was reached.All items which were not coded "adequate" were reviewed to determine the source of the problem and, if relevant, their need for adaptation.The decision to modify or not modify the item was based on the assessment of the impact of the problem, clinical judgment, and Rasch analysis results of a previous study [14].Items were modified by two study investigators.

Participants
Based on the reliability data from a previous study [8], a sample size of 8 participants was determined to provide sufficient power to assess our a priori hypothesis that the reliability of the WheelCon-M-F-P would exceed an intraclass correlation coefficient of 0.80, given an α (Type I error) of 0.05 and β of 0.80 [25].To assess the validity of the WheelCon-M-F-P in relation to the WST-Q and the COSA, a sample size of 20 was estimated based on variability data from the same study [8] (α [Type I error] = 0.05 and β = 0.80) [26].To accommodate potential study dropouts, the sample size was conservatively adjusted by 20% for a total sample size of 24 participants.

Recruitment and screening
Recruitment followed the same process as in Phase 2, plus additional recruitment strategies, including study advertisements posted at the CRME and on social media, such as Facebook.The inclusion criteria were the same, except the location was restricted to Montreal for this phase of the study.

Procedure
For this test-retest methodological study, two data collection sessions were conducted with each participant.Prior to the first session, the participant, their parent(s), or their occupational therapist completed a socio-demographic questionnaire which collected information including age, diagnosis, years of wheelchair experience, and level of help needed in daily activities.In the first session, which took place at either École Victor-Doré or École Joseph-Charbonneau, the participant completed the WheelCon-M-F-P, the WST-Q, and the COSA.Approximately one week later, the WheelCon-M-F-P was re-administered.This second session took place at the same location as the first meeting for all participants, apart from one which was held via phone for the convenience of the participant.For each questionnaire, the investigators explained the purpose of the measure, instructions for completion, and the response scale.

WheelCon-M-F-P
The WheelCon-M-F-P used in this Phase is the final version (see Supplementary Material) as developed in Phase 1 and Phase 2 of this study.The outcome measure has 33 items in the areas of negotiating the physical environment (15), performing activities in the manual wheelchair ( 6), knowledge and problem-solving (4), advocacy (3), managing social situations (3) and managing emotions (2).For the 33 items in this paediatric version, the stem is "as of now, how confident are you…".The response scale is based on a 5-level Smiley Face Likert scale (red face; tan face; yellow face; aqua face; and green face), with descriptive anchors for the red face (not confident) and green face (very confident).In terms of administration, the evaluator explained to the children how to complete the questionnaire, its purpose, and the response scale.
Children were asked to mark the face that they felt most closely depicted their confidence level with each item at that moment.The study investigator stayed next to the children to clarify items, answer questions, read the items for the child (if needed), or mark the face on the paper if it was not possible for the child to do so.To score the WheelCon-M-F-P, each response was first transformed to a number (red face = 0, tan face = 1, yellow face = 2, light green face = 3, green face = 4) and then the average questionnaire total score was calculated by adding the scores of each question and dividing it by the number of questions answered.The questionnaire total score was transformed into percentages by dividing it by four (maximum possible score) and then multiplying it by one hundred as follows: ((sum of each item score)/ ([number of items] × 4)) × 100.

Comparison questionnaires
The Wheelchair Skills Test Questionnaire 5.0, French version, is a self-administered questionnaire that measures wheelchair skill capacity, confidence, and performance in manual wheelchair users.This questionnaire includes 33 items and the response scale is based on three separate Likert scales of four levels for capacity (i.e."I can safely do the skills, by myself, without any difficulty," "Yes, but not as well as I would like," "I have never done the skill or I do not feel that I could do it right now," "My wheelchair does not have the parts to allow this skill"), confidence (i.e.very confident, fairly confident, somewhat confident, not confident, not possible with this wheelchair) and performance (i.e.daily, weekly, monthly, yearly, never, not possible).The measurement properties of the WST-Q version 2.4 were very good to excellent [27].Moreover, this questionnaire has been chosen to verify the validity of the WheelCon-M-F-P because WST-Q version 4.3.3 was recently demonstrated as valid among PMWUs [28].Furthermore, it addresses the main domain of the WheelCon-M-F-P, more specifically, negotiating the physical environment in a manual wheelchair.
The COSA, the French version, is a self-administered questionnaire that measures how well children feel they competently fulfil expectations and the responsibilities associated with day-to-day activities and the importance they attach to these activities [29].This questionnaire includes 25 items rated using a 4-point Likert scale for ability (i.e."I have a big problem doing this," "I have a little problem doing this," "I do this ok," and "I am really good at doing this") and perceived importance (i.e."not really important to me," important to me," "really important to me," and "most important of all to me") for a variety of tasks, such as "keeping my body clean," and "have enough time to do things I like."The final score for ability was calculated using the method of the WST-Q composite scores since the COSA does not provide quantitative scoring.We attributed numerical scores (0-3) to the scale (by attributing scores from 0 to 3 for each item).The measurement properties (e.g.content, structural, and substantive validity) of this questionnaire are very good to excellent [29].This questionnaire was selected as a comparison because it addresses an important area of the WheelCon-M-F-P, specifically, activities of daily living.

Data analysis
Descriptive statistics were calculated to describe the study population and the scores of the three questionnaires (WheelCon-M-F-P, WST-Q, and COSA).The normality of the data was calculated using the Shapiro-Wilks test.Cronbach's alpha was used to calculate the internal consistency of the WheelCon-M-F-P.The test-retest reliability of the WheelCon-M-F-P was measured with the intraclass correlation coefficient (ICC 1.1 ).Responsiveness was calculated from standard error of measurement (SEM) and the smallest real difference.The Bland-Altman limits of agreement plot was used to provide a visual assessment of how individual WheelCon-M-F-P scores varied between baseline and follow-up.To measure the validity of the tool in relation to other measures, Pearson's correlation was used to calculate the relationships between the WheelCon-M-F-P, the COSA, and the WST-Q, as well as participants' age, sex, and years of wheelchair experience.The strength of the correlations were interpreted whereby 0.90-1.00 is very high, 0.70-0.90high, 0.50-0.70moderate, 0.30-0.50low and 0.00-0.30negligible [30].IBM SPSS Statistics version 25 was used to conduct all analyses.

Phase 1: item and scale adaptation
Figure 2 depicts the flow of the item adaptation phase.Of the 65 WheelCon-M-F items, 35 were removed, 30 were deemed relevant for the paediatric population, of which 25 required adaptation and 5 did not.In addition, 6 items were added.The example items described below have been translated from French to English by a bilingual member of our team.
28 of the 35 removed items were deemed not relevant as they did not represent confidence-challenging situations for PMWUs, such as "… how confident are you to present yourself as you wish to be seen while in your wheelchair when you want to impress others, such as during a job interview?"and "… how confident are you to advocate for changes in your community, such as having a curb cut added in your neighbourhood to improve your accessibility?" Three items were removed because they were replaced with another item.For example, the item "… how confident are you figuring out how to negotiate a challenging, and unusual physical obstacle?"was replaced with "… how confident are you that you can find a solution when you have a problem, like a physical obstacle?"The other four items were removed because they represented confidence in challenging situations of combined skills, which was deemed too complex for children, such as "… how confident are you moving your wheelchair through 5 cm (2″) snow then up a curb cut?" Of the 25 items requiring adaptation, 4 items were modified to be more relevant to children's daily activities, such as changing the item "… how confident are you presenting yourself as you wish to be seen while in your wheelchair around acquaintances, colleagues, or peers?" to "…how confident are you feeling comfortable while in your wheelchair around people you know (friends, family)?"The remaining 21 items were modified to simplify sentence structure or terminology for ease of comprehension.For example, the item "… how confident are you to open, go through, and then close a standard 81 cm (32″) lightweight door?" was modified to "…how confident are you to open a standard door, go through it, then close it?" In addition to item adaptation, there were 6 new items added to the WheelCon-M-F-P.Two of these items were added to reflect confidence in manual wheelchair use at school, such as "…how confident are you to carry school materials while in your wheelchair?"The four others were added to reflect confidence in situations that are part of children's everyday life, such as "… how confident are you to play with other kids who aren't in a wheelchair?" To enhance the clarity for children, the WheelCon-M-F, paediatric version instructions were adapted, as well as the pronoun used in the stem and items, and the response scale.Specifically, the instructions were simplified.For instance, the sentence "For example, a person may be 82% confident they can memorize a grocery list of 5 items, but only 63% confident they can memorize a grocery list with 10 items."was adapted to "For example, a person may be pretty confident to read a five-page book, but not confident enough to read a hundred-page book."In terms of the pronoun, the use of the second person singular was changed to the use of the first person singular.Moreover, we adapted the questionnaire by changing the response scale from a Likert scale (0-100, where 0 represents Not confident and 10 represents Completely confident, increments of 10) to a 5-Level Smiley Face Likert scale that included five faces, as described above.

Phase 2: item refinement
The 4 participants were aged between 14 and 18 years old, 3 were female and their diagnoses were either cerebral palsy (n = 3) or osteogenesis imperfecta (n = 1).Participants' demographic data are presented in Table 1.
Of the 36 WheelCon-M-F-P items resulting from Phase 1, 20 (55.6%) were coded as "adequate response" and 16 (44.4%)were coded as "clarification", "qualified", or "inadequate".None of the participants' responses were coded "refusal to answer", but some responses were made without thinking aloud.Table 2 describes each WheelCon-M-F item, its refinement (if applicable), and the number of behavioural codes per item.The items described in this table were translated from French to English for ease of read.
Ultimately, 14 items were modified to enhance the correct interpretation of the item.The following example items provided were also translated from French to English.Items modified by adding details (n = 11) included "I am confident that I can move my wheelchair over carpet" was refined to "I am confident that I can move my wheelchair over thin carpet".One item was modified by removing detail (n = 1), which was the removal of the word "standard" in the item "I'm confident that I can open a standard light door, go through it, and then close it."Some items were modified by replacing words (n = 2), such as "I am confident that I can go up a curb cut with my wheelchair" was refined to "I am confident that I can move my wheelchair up a curb cut".Furthermore, three items were removed as they were deemed not relevant or too difficult to understand for children, such as the item "I am confident that I can know what my wheelchair can and can't do".The 20 items with the code "adequate" remained the same, except for one.This item was changed for consistency with the terminology of another item, which was the item "I am confident that I can go down a curb cut with my wheelchair" was refined to "I am confident to go down a curb cut with my wheelchair".

Participants
The sociodemographic and wheelchair-use characteristics of the 22 participants are presented in Table 3.The average age of the sample was 12.6 ± 3.3 years with a range of 7 to 18 years old.There were equal representations of girls and boys in the sample.The majority (48%, n = 21) of the sample had a diagnosis of cerebral palsy.All participants went to a specialized school.The majority (72.7%, n = 22) were native Canadian French-speakers.All participants lived at home with their parents.Only one participant dropped out secondary to illness.Move my wheelchair over a threshold, such as between rooms.Move my wheelchair over a threshold (for example, between two doors).
Use my wheelchair in small spaces, (for example, in a bathroom) Prepare myself a simple meal while using my manual wheelchair.Move my wheelchair up a small ramp.Go down a slanted sidewalk with my wheelchair.

1
Move my wheelchair down a curb cut.
Move my wheelchair over a drainage gate.

Item deleted
Roll on snow.Move my wheelchair in a small amount of snow.
Use my manual wheelchair to go push the crosswalk button and cross the street before the light turns red.Use my manual wheelchair to go push the crosswalk button and cross the street before the traffic light turns red.
Cross a street without traffic lights and where traffic is light.
Cross a street without traffic lights and where there are not many cars.
Move with my wheelchair for a short distance over ground that is flat and dry and has short grass.Move with my wheelchair over an uneven and cracked surface (for example, over a cracked sidewalk).Use strategies (for example, making people laugh), to help people feel comfortable if they are not sure how to act because i use a wheelchair.

no change
Correct people when they are wrong about me when i use my manual wheelchair.

no change
Feel comfortable when i am in my wheelchair with people i know (friends, family).

no change
Feel comfortable when i'm in my wheelchair with people i do not know.

no change
Find a solution when i am faced with a problem, (for example, finding a way around an obstacle that prevents me from passing).

Item deleted
ask for changes that i would like make to my manual wheelchair, like a different and more comfortable cushion.ask for changes that i would like make to my school, like a change in the bathroom set-up.a: adequate (participant gives an answer that meets item objectives); C: clarification (participant asks for clarification of item, or makes statement indicating uncertainty about item meaning); Q: qualification (participant gives an answer that meets item objective but is qualified to indicate uncertainty about accuracy); i: inadequate (participant gives an answer that does not meet item objective).
For the purpose of this article, we translated the items from Frech-Canadian to english. the WheelCon-M-F-P French-Canadian can be found in the supplementary Materials.
a Changes are bolded.

WheelCon-M-F-P scores
In the WheelCon-M-F-P administration, three participants did not clearly answer a total of seven items (e.g. they gave multiple answers or no answer).These ambiguous responses were omitted from the calculation of the total scores for two reasons: there was no clear indication given to the participant that each item required only one answer, and the study investigators did not catch these response errors during data collection.The scores of the WheelCon-M-F-P demonstrated a normal distribution at baseline (Shapiro-Wilk p = 0.135).The mean (SD) of the WheelCon-M-F-P score was 73. 30 ±14.29 at baseline and 73.29 ±16 16 .at follow-up.

Internal consistency
Cronbach's ɑ coefficient was 0.846 for the 33-item WheelCon-M-F-P (at baseline), indicating that the outcome measure has internal consistency.

Reliability
The 1-week retest ICC 1,1 of 0.82 (95% CI 0.568-0.924)supported our a priori hypothesis.The Bland-Altman plot (Figure 3) analysis indicated that the 95% limits of agreement between the test and re-test ranged from −24.43 to 24.45.The distribution of values is mainly below the mean difference, demonstrating a systematic bias towards a lower score.No ceiling or floor effect was detected and there were only two outliers (i.e.9% of the sample) beyond the upper limit of agreement [31].

Responsiveness
The standard error of measurement was 3.05 and the smallest real difference was 8.45, providing an indication of the minimal change in score that would reflect a meaningful change beyond measurement error for a group of individuals and a single individual respectively.

Validity with other variables
Correlations between the WheelCon-M-F-P (Time 1) and the WST-Q and the COSA were statistically significant in the direction and of the magnitude hypothesized.However, correlations with age and experience were not statistically significant, nor were they in the hypothesized direction.Table 4 presents all correlations.

Feedback from participants
During the administration of the WheelCon-M-F-P, certain word combinations were commonly asked to be clarified by participants, such as: "pente légère" (small incline), "surface inégale et crevassée" (uneven and cracked surface) and "Gravier" (gravel).Some also had difficulty knowing where to write their answers on the printed version of the questionnaire, resulting in incomplete or ambiguous responses.

Discussion
In this study, the WheelCon-M-F was adapted into the WheelCon-M-F-P resulting in a 33-item self-report outcome measure designed to assess confidence with manual wheelchair use among children.By analysing the secondary data in Phase 1, we were able to modify the WheelCon-M to highlight the situations that challenge children's confidence while using manual wheelchairs.These modifications are further justified by the premise discussed in Pituch et al. [19] that PMWUs are often assisted by a parent or a caregiver and are not often left alone, rendering specific situations in the WheelCon-M not pertinent.Consequently, the removal of 35 WheelCon-M items and the addition of 6 items result in an outcome measure with items deemed particularly confidence challenging for PMWUs.Specifically, items in the domain of negotiating obstacles were removed considering that PMWUs are in the presence of a parent or an adult able to assist them with obstacles such as high curbs, steep inclines, or getting over a gap.Other items in the domain of performing activities in the manual wheelchair were added to the WheelCon-M-F paediatric version to be more inclusive of the PMWUs community environment (i.e.playgrounds and school) while others were removed since the environment does not pertain to children (i.e.items including the workplace).Also, 25 items were modified to improve their clarity.In general, for those items, we simplified the sentence structure or terminology, and we removed measurements details.
In Phase 2, the think-aloud process evaluated if each item was interpreted as intended by PMWUs.Most of the items (56%, n = 20) were correctly interpreted by the participants, demonstrating that others were still misinterpreted or irrelevant.The items being misinterpreted by PMWUs were often abstract situations which were then clarified with concrete examples.Irrelevant items did not correspond to the reality faced by children so those were completely removed.Although our Phases 1 and 2 mirrored the methods used to adapt the adult WheelCon-M into a version for power wheelchair users [13], the participants in Phase 3 demonstrated that some words remained difficult for PMWUs to understand.Perhaps the challenging vocabulary would have been addressed during Phase 2 had there been a larger, more heterogeneous sample that better represented children of various ages, diagnoses, and education levels.Furthermore, it is worthwhile questioning whether the scoring method used in the WheelCon-MF-P was the most optimal.It may have been better to have a scoring method with a description for each face, rather than only anchoring descriptions for the red and green faces.We have subsequently added a description for each face (Supplementary Material).
The results from the evaluation of the measurement properties of Phase 3 show internal consistency between the 33 items, responsiveness, test-retest reliability, and validity in terms of the WheelCon-M-F-P with the WST-Q and COSA, which is comparable to the results obtained for the adult version [8], the Italian version [16], the French-Canadian version [17] and the short version [14].While our hypotheses regarding the correlations of WheelCon-M-F-P scores with the WST-Q and COSA were supported, those regarding age and experience were not in terms of strength and direction.We formed our hypothesis about age based on the fact that young children often demonstrate higher levels of confidence, but actually, the average age of our participants coincided with the well-documented downward trend for confidence level for both boys and girls which starts around the ages of 9-12 [32].With regards to experience, one potential reason for the negative and non-statistically significant correlation between the number of years of wheelchair experience and wheelchair confidence could be that confidence is task-dependent, as described in the Social Cognitive Theory [33].In other words, while paediatric wheelchair users do gain experience for specific tasks, they will eventually face new and more complex ones for which they do not have the necessary skills, and, as a result, their confidence diminishes [18].Furthermore, it has been shown that receiving more assistance with wheelchair use is associated with weaker confidence, whether the support provided is at a physical [9] or emotional [8] level.Therefore, seeing that PMWUs often require assistance with wheelchair use and are often accompanied [19] and that only 18.2% of the participants (n = 4) in the current study required "no assistance" (supervision, verbal, or physical support), lower wheelchair confidence scores despite a number of years of experience is better explained.Interestingly, the Bland-Altman analysis showed that there was a slight systematic bias towards lower scores.This result suggests that the outcome measure contained items with which the PMWUs may have experimented between the two times of administration, resulting in slightly higher scores at the re-test.Furthermore, that only 4 participants (18.2%) did not require any type of assistance for manual wheelchair use may have biased the perception of the task difficulty for each item in our questionnaire.Corroborating this idea is the fact that our sample population all attended specialized schools.

Limitations
Some limitations of this study are important to note in planning for future research and interpreting the current results.Participants in all study phases attended specialized schools adapted for manual wheelchair use, in one city, and from one rehabilitation center and its affiliated schools, and thus face similar confidence-challenging situations which limits the generalizability of the results.In Phase 2, the small sample size between the ages of 14 and 18 may have resulted in a misrepresentation of younger children's perspectives.Despite the explicit explanation of the think-aloud process and the study investigators' best efforts to remain objective, social desirability bias may have occurred.Finally, although 27% of the participants in Phase 3 were not native Canadian French speakers, which may be perceived as an explanation for the need for clarification of certain terms, all participants attended francophone schools and thus we do not believe that language created a challenge in comprehension.
Future directions will include the preparation of the outcome measure in larger font, increased spacing between each item, and clear instruction on how to indicate the face best corresponding to the level of confidence for each item (circling, marking with an "x", or using a stamp).We also recommend adding a sample question/answer to demonstrate how to proceed in completing the outcome measure to ensure that the instructions are well understood by the PMWU.Finally, there are plans to develop a mobile application featuring this questionnaire and the other versions of WheelCon to facilitate their use.Further evaluation of the WheelCon-M-F-P is warranted given the changes to the response scale, administration instructions, and formatting.

Conclusion
The WheelCon-M-F-P is a new 33-item self-report outcome measure with preliminary evidence of its validity based on response processes, internal structure, and relations with other variables.In practice, this questionnaire may be used after a child receives a manual wheelchair to examine areas of low confidence so that a targeted confidence-enhancing intervention may be provided.It may also be used as an outcome measure to evaluate the effectiveness of interventions.

Figure 1 .
Figure 1.sequential phases of the WheelCon-M-F adaptation into a paediatric version.

Figure 2 .
Figure 2. Results of item adaptation in Phase 2.

2 2 1 2
Prepare a simple meal (for example, toast or cereal) in my kitchen while using my manual wheelchair.open a light standard door, go through, and then close it.1 open a light door, go through it, and close it.

2 1
with my manual wheelchair over a flat gravel surface for a short distance.outside (for example, in the schoolyard or at the park) with my wheelchair.Carry school supplies when i use my wheelchair at school. 1 Carry my lunch box and a book in my backpack when i use my wheelchair at school.Participate in recreation and sports in my wheelchair.
to move with my manual wheelchair in a situation that makes me anxious or nervous.to someone how to move my manual wheelchair if it is stuck.help when i do not need it (for example, if someone offers to push my wheelchair down the street, but I do not need help).Know what to do if i fall out of my wheelchair.

2 1 1
ask for a change to my manual wheelchair if I need it (for example, a more comfortable cushion).
to be made at my school (for example, a change in the set-up of the classroom or bathroom).

Table 2 .
P think aloud WheelCon-M-F-P results and item refinement.

Table 4 .
Correlations between the WheelCon-M-F-P with the Wst-Q, Cosa, age and wheelchair experience.