Mobility and Quality of Life: A Cross-Sectional Psychometric Evaluation of the Validity and Reliability of a Dutch Translation of the MobQoL-7D Outcome Measure

Abstract

Background: The Mobility and Quality of Life-7 Dimension (MobQoL-7D) is a new patient-reported outcome measure for mobility-related quality of life. Our aim was to translate and test a Dutch-language version. Methods: A cross-sectional psychometric evaluation study was undertaken. The sampling frame was community-dwelling adults living in the Netherlands who had a long-term (>6 months) mobility impairment. Participants were recruited through a Dutch research agency, and data were collected via online survey. Statistical and psychometric analyses were undertaken to assess the interpretability, validity and reliability of the MobQoL-7D Dutch, including assessment of missing data, floor/ceiling effects, test–retest reliability, structural validity, known-group validity and convergent validity. Results: n = 308 respondents completed the survey; sub-group sample sizes ranged from n = 29 to n = 87. No issues with missing data were found. Despite ceiling effects per item (ranging from 23.1% to 56.5%), there were no floor/ceiling effects for overall index values (12.3% and 0%, respectively). The findings show excellent test–retest reliability of the index value over a two-week period (n = 37; ICC = 0.95), and potential discriminative ability to detect differences between known groups. Factor analyses confirmed unidimensionality. Conclusions: The results provide promising evidence of the validity and reliability of the MobQoL-7D Dutch; further research is needed to confirm these findings.

1. Introduction

Worldwide an estimated 65 million people use some form of mobility aid, such as a wheelchair, walking stick or walking frame [1]. For people with a mobility impairment, mobility aids are an essential tool for moving in and outside of the home, performing daily activities, participating in everyday life, and maintaining independence.

In the Netherlands, approximately 14% of the 17.5 million citizens use a mobility aid for daily mobility [2]. Wheelchairs are one of the most commonly supplied mobility aids in the Netherlands, accounting for around 40% of all mobility aids. According to the World Health Organization (WHO), the term “wheelchair” has a broad definition and is “inclusive of all types of wheeled mobility devices including manual, power, power assisted, wheelchairs, tricycles and scooters, and essential components or features such as wheelchair cushions and postural supports” [1]. Wheelchairs and other similar mobility aids provide direct access to mobility, while secondary aids like hinges or custom comfort chairs are secondary to mobility but are often a prerequisite or facilitator to a primary mobility aid. In the Dutch care and rehabilitation context, an extended definition of tricycles is applicable. Special bikes like tandems, duo-bikes and handbikes are also qualified as wheeled mobility devices for outdoor mobility.

Provision of mobility aids in the Netherlands is governed by four basic healthcare-related acts: the Health Insurance Act (HIA) [3], the Long-Term Care Act (LTCA) [4], the Social Support Act (SSA) [5] and the Employee Insurance Agency (EIA) [6]. Both the SSA and the LTCA provide support in the form of mobility aids like wheelchairs for people living at home (SSA) or in an institution for long-term care (LTCA). Mobility aids provided through the EIA are primarily assigned for specific work or school activities. Mobility aid provision through these four acts varies in base and execution.

In 2022, 350,000 mobility aids were provided through the SSA for use by an estimated 250,000 people [2]; 70–75% of these individuals use one mobility aid, 15–20% use two mobility aids and the remaining 10–15% have three or more mobility aids in use [7]. The 350,000 mobility aids include the following:

• Manual wheelchairs;
• Power wheelchairs (including power assist devices);
• Mobility scooters;
• Transfer aids (standing device, active, passive, and stairlift);
• Bicycles (including tricycles, handbikes and special bikes);
• Other (toilet/shower chairs, custom moulded seating supports, comfort chairs, disability chairs, and wheelchair cushions).

To provide people with mobility aids, the SSA is governed to collaborate with providers and suppliers through public tender processes in line with European Procurement Laws. In these tenders, the price and quality of the mobility aid provision are assessed to choose the most appropriate provider per municipality (or group of municipalities). In the last few years, the pricing has been declining, while the demands of the end-user are increasing due to technological improvements, more complex disabilities and a growing population of people who require mobility aids.

In light of lower pricing trends across the sector, it is important to shift the tone of mobility aid provision away from primarily focusing on cost. Instead, there needs to be a focus on measuring the benefits for the end user in terms of autonomy, independence, inclusion and prevention of extra care needs affecting quality of life. Shifting the perspective also demands a baseline measure to determine the health outcomes of mobility aid users and, in turn, to focus on patient-reported outcomes and broader measures such as quality-adjusted life years (QALYs).

Dutch versions of generic health-related quality of life instruments, including the EuroQol 5-Dimension 5-Level (EQ-5D-5L) [8], the Short Form-36 (SF-36) [9], and the World Health Organization Quality of Life instrument (WHOQOL-BREF) [10], have been utilised with Dutch patient groups. However, these measures do not explicitly focus on the multifaceted relationship between mobility and quality of life. Furthermore, a previous review found that generic preference-based health-related quality of life measures, such as the EQ-5D-5L and SF-6D (derived from the SF-36), can lack sensitivity in mobility-impaired populations [11], and both the WHOQOL-BREF and EQ-5D-5L contain only one dimension specifically related to mobility.

Other validated Dutch instruments are available which focus more explicitly on impairment, physical functioning and rehabilitation, including the Canadian Occupational Performance Measure (COPM) [12], the Dutch version of the Quebec User Evaluation of Satisfaction with Assistive Technology (D-QUEST) [13], the Utrecht Scale for Evaluation of Rehabilitation–Participation (USER-P) [14], and the Impact on Participation and Autonomy Questionnaire (IPAQ) [15]. However, none are specifically focused on the concept of mobility-related quality of life. We therefore see a key gap in the literature for a validated Dutch outcome measure which provides a comprehensive, mobility-specific assessment of quality of life, capturing physical, emotional, and social domains.

The Mobility and Quality of Life-7 Dimension (MobQoL-7D) is a novel patient-reported outcome measure for mobility-related quality of life [16] (see Supplementary File S1). Originally developed in English, the MobQoL-7D was designed for use in both clinical practice and research to evaluate the effectiveness of mobility aids in improving mobility-related health and quality of life outcomes. The MobQoL-7D health state classification system contains seven items, each representing a distinct conceptual dimension of mobility-related quality of life; these are Accessibility; Contribution; Pain and discomfort; Independence; Self-esteem; Mood and emotions; and Anxiety. The original English version has been psychometrically validated and contains two underlying factors within the item structure: (1) physical and role functioning related to mobility, and (2) mental well-being related to mobility. A preference-based scoring system has also been developed for the UK population, demonstrating how QALYs can be calculated using MobQoL-7D data [17].

We believe that the MobQoL-7D could offer a unique approach to measuring mobility-related quality of life in the Dutch population. Due to the unique cultural and societal characteristics of the Netherlands, we have designed a programme of research to translate and test a Dutch version of the MobQoL-7D (known as MobQoL-7D Dutch).

Aims and Objectives

The key objective of this study was to translate the MobQoL-7D into Dutch and assess the basic measurement properties, interpretability and psychometric properties of the translated measure in the Dutch population. The following question was explored: Is the MobQoL-7D Dutch a valid and reliable measure of mobility-related quality of life in the Dutch population?

2. Materials and Methods

A cross-sectional psychometric evaluation study was undertaken. Quantitative data were collected through online surveys. The methodology followed a similar procedure used for the development of the original version of the MobQoL-7D outcome measure [16] and was influenced by COSMIN guidelines [18]. A wide range of measurement properties were examined to assess the validity and reliability of the MobQoL-7D Dutch. Validity refers to the extent to which a measurement tool is truly measuring the intended underlying construct or latent variable—in this case, mobility-related quality of life. Reliability refers to measurement error, including repeatability and consistency.

2.1. Data Collection and Outcome Measures

The primary sources of data were online questionnaire surveys distributed to Dutch-speaking people in the Netherlands who had a long-term (>6 months) mobility impairment and used at least one mobility aid. The questionnaire survey contained the following:

• Demographic and personal characteristics including gender, age, employment status, medical diagnosis and perceived stability of health.
• Information about the type of mobility aid used, and duration/frequency of use.
• MobQoL-7D Dutch translation.
• Additional outcome measures (translated into Dutch):

  ◦

  Unvalidated translation of the EQ-5D-5L;

  ◦

  Unvalidated translation of the ICEpop CAPability measure for Adults (ICECAP-A);

  ◦

  Unvalidated translation of the Psychosocial Impact of Assistive Devices Scale (PIADS).

The three additional outcome measures were included in the survey to allow preliminary assessment of convergent validity and to define groups of respondents; please see the “Hypothesis testing” section below for a full description of how the groups were defined using these measures. Due to using unvalidated translations of the chosen outcome measures, equivalence to the validated versions cannot be assumed; our approach to translation is explained in the “Translations of measures” section.

The EQ-5D-5L [19] is widely used internationally to calculate QALYs. Respondents are asked to rate their health by scoring each of the five dimensions using one of five response options, ranging from no problems to extreme problems. The five dimensions are Mobility; Self-care; Usual activities; Pain/discomfort; and Anxiety/depression. Health states are converted to a single summary index value by weighting the level of each dimension and deducting those weights from 1 (perfect health). The measurement scale ranges from 0 (death) to 1 (perfect health); health states may be valued as worse than death, depending on the value set used.

The ICECAP-A is a measure of capability and well-being [20], which is suitable for measuring and valuing effects of non-health interventions, such as social care [21]. The ICECAP-A offers an alternative approach to the EQ-5D-5L for measuring generic health outcomes in certain types of economic evaluations. The ICECAP-A can be used to calculate an estimated cost per “year of full capability”, a theoretical alternative to the QALY approach [22]. The ICECAP-A comprises five dimensions: Feeling settled and secure; Love, friendship and support; Being independent; Achievement and progress; and Enjoyment and pleasure. Each dimension has four response choices; for instance, the response choices for the “Being independent” dimension range from “I am able to be completely independent” to “I am unable to be at all independent”. The measurement scale ranges from 0 (no capability) to 1 (full capability).

The PIADS is an outcome measure for assessing the effects of assistive technology on functional independence, well-being and quality of life [23]. The questionnaire contains 26 separate items, covering concepts such as happiness, independence, efficiency and productivity. For each item the respondent is asked to indicate the extent to which their assistive technology decreases (−3, −2, −1) or increases (+1, +2, +3) a particular concept. Individual item scores are subsequently grouped to inform three averaged sub-scale scores in three domains: Competence; Adaptability; and Self-esteem. The PIADS is considered to be generically applicable across all major forms of assistive technology.

Participants were also asked to report whether they perceived their health to be stable or unstable, using a single self-rated health question (i.e., “Overall, do you consider your health to be stable at the moment?”). A follow-up “retest” survey with ~10% of the sample was undertaken two weeks after the initial survey. The retest data were analysed in a preliminary evaluation of the test–retest reliability of the MobQoL-7D Dutch, based on the assumption that repeated measurement of a reliable outcome measure should show no substantial changes over a short time interval.

2.2. Translations of Measures

The Dutch language translation of the MobQoL-7D was developed by the research team prior to the initiation of this research. Firstly, the MobQoL-7D was translated into Dutch by a professional Dutch language translator; secondly, the initial translation was checked for linguistic and cultural accuracy by 2 native Dutch speakers and then back-translated. Amendments were then made, and the final version of the MobQoL-7D Dutch was confirmed (see Supplementary File S2).

Despite existing Dutch translations of the other included outcome measures, we opted to produce our own unvalidated Dutch translations to ensure linguistic consistency across the data collection tools. Although we did not identify any specific issues with the validity of the original translations of the EQ-5D-5L, ICECAP-A and PIADS, we produced our own unvalidated translations in accordance with international plain-language guidance [24] and the Common European Framework of Reference for Languages [25] to ensure all of the outcome measures were consistent and accessible for individuals with varying health literacy levels. Similarly to the MobQoL-7D Dutch translation, the outcome measures were forward translated by a native Dutch speaker from the original English language version and crosschecked by another native Dutch speaker.

We subsequently compared our unvalidated translations to the official Dutch translations of each outcome measure; although some minor variations were found, they were generally consistent. Total equivalence between our translations and the official translations cannot be assumed; therefore, convergent validity results should be considered with caution. The translations of the outcome measures can be viewed here: https://doi.org/10.6084/m9.figshare.31076071.

2.3. Recruitment and Sampling

Participants were identified and recruited through a customer panel convened by the research agency ZorgFocuz [26], which specialises in post-market surveillance in healthcare, welfare and social domains in the Netherlands. Recruitment was focused on mobility-aid provision for community-dwelling adults. Participants were identified by screening for individuals who had recently received a mobility aid through the SSA. Potential participants were sent a study information sheet and an invitation to the study survey via email by ZorgFocuz.

Sampling focused on identifying individuals who used a mobility aid and had a long-term mobility impairment, as these individuals represent the primary target population for the MobQoL-7D (and all translated versions of the MobQoL-7D). For the purpose of this research, a “long-term” mobility impairment was defined as any condition, impairment, disability or illness causing impairment to mobility for 6 months or longer. Furthermore, a “significant” mobility impairment was defined as any impairment to mobility which necessitates the use of a mobility aid and/or a mobility-enhancing intervention to enhance, maintain or facilitate mobility or to reduce complications related to mobility impairments.

Maximum variation sampling was utilised to create a diverse sample of individuals with a wide range of mobility impairments and experience in using a variety of different mobility aids. Formal sample size calculation was not undertaken; we instead aimed to exceed the stated guidance of a minimum sample size of n = 250 for quantitative instrument validation in patient groups [27] and n = 50 for hypothesis testing in instrument validation [18].

2.4. Inclusion and Exclusion Criteria

The following broad inclusion criteria were used:

• Aged 18 years or over;
• Have a long-term (i.e., lasting or expected to last >6 months) impairment, injury or condition which necessitates the use of any form of mobility aid to improve or facilitate independent mobility;
• Capacity to provide informed consent;
• Ability to communicate in Dutch (at least Level B1 according to the Common European Framework of Reference for Languages [25]).

Specific disabilities, conditions, or functional status were not explicitly targeted, as mobility aids are used by people with a vast array of conditions, impairments and injuries.

2.5. Data Handling and Ethical Approval

Data protection regulations and guidelines were adhered to. To safeguard respondent privacy, all data were anonymised, and each participant was assigned a unique, non-identifiable ID. Ethical approval was granted by an independent ethics committee (METC Brabant; NW2024-50).

2.6. Analyses

An array of analyses was undertaken using SPSS^® (v29.0) and JASP (v0.95.4). Planned methods of data analysis were divided into three stages: (1) basic measurement properties, interpretability and reliability; (2) structural validity; and (3) hypothesis testing.

2.6.1. Basic Measurement Properties, Interpretability and Reliability

Both the individual item (i.e., individual survey question) scores and overall index values (i.e., converted scores utilising responses to all items) of the MobQoL-7D Dutch were examined to assess various measurement properties and psychometric properties of the Dutch translation. The overall index value was calculated using the UK value set (available here: https://cheme.bangor.ac.uk/mobqol (accessed on 13 June 2024); at present a value set for the Netherlands is not available. This scoring system produces a summary score between 0 (death/worst possible health state) and 1 (best possible health state).

The following criteria were examined:

• Missing data: The threshold for problematic measurement was set at >4% missing data per item. High levels of missing data can indicate that respondents find a tool to be too difficult, confusing, long or intrusive.
• Floor/ceiling effects: Floor and ceiling effects can indicate a lack of sensitivity in the measurement scale. For overall index values, a ceiling effect was defined as ≥15% of respondents reporting the best possible health state (i.e., choosing level 1 for each item; 1111111); and a floor effect was defined as ≥15% of respondents reporting the worst possible health state (i.e., choosing level 4 for each item; 4444444). COSMIN guidance does not specifically state a threshold for floor/ceiling effects; however, ≥15% is generally considered to be an appropriate threshold [28]. Although floor/ceiling effects are principally evaluated at the scale level (i.e., total score), in the interest of transparency, we also assessed floor/ceiling effects for the individual items. The threshold was again set at ≥15% of responses on the first (level 1; ceiling) or last (level 4; floor) response choice for each item. Level 1 represents the ‘best’ response choice (i.e., no problems) for each item and is thus the ‘ceiling’; likewise, Level 4 represents the ‘worst’ response choice (i.e., extreme problems) for each item and is thus the ‘floor’.
• Test–retest reliability: This provides an indication of whether a tool is reliably and consistently measuring a construct over time. As per COSMIN guidance [29], test–retest reliability was assessed using Intraclass Correlation Coefficient (ICC) (two-way random effects model, based on absolute agreement) for the overall index value and weighted Cohen’s kappa (using quadratic weights) for the individual items. ICC values between 0.75 and 0.90 indicate good reliability, and values greater than 0.90 indicate excellent reliability [30]. A minimum kappa value of ≥0.60 is the recommend threshold for adequate agreement between measurements [31].

2.6.2. Structural Validity

• Exploratory factor analysis (EFA): As the MobQoL-7D Dutch is a novel PROM translation, EFA was undertaken to examine the inter-relationship between items and underlying factorial structure [32]. Due to the ordinal nature of the 4-point Likert-type scale utilised in the MobQoL-7D Dutch, EFA was performed using Principal Axis Factoring and with polychoric correlation. Rotation was not required, as only one factor was identified. Sampling adequacy was assessed using the Kaiser–Meyer–Olkin (KMO) test (KMO ≥ 0.90 indicates excellent sample adequacy), and Bartlett’s test of sphericity was used to test the adequacy of correlations between items (p < 0.05 confirms data suitability for EFA) [33]. The number of independent factors was determined by eigenvalues ≥ 1 [33] and interpretation of the screeplot. As per COSMIN guidance, relevant fit indices included factor loadings for each item and variance across the items explained by the identified factor (COSMIN thresholds: factor loadings ≥ 0.30 per item and explained variance ≥ 50%, respectively) [29].
• Confirmatory factor analysis (CFA): This was undertaken to further validate the factorial structure identified by the EFA and to also examine potential factorial fit with the two-factor structure of the original MobQoL-7D (factor 1: physical/role function related to mobility; and factor 2: mental well-being related to mobility). The CFA models were estimated using weighted least square mean and variance (WLSMV), which is considered most appropriate for ordinal data (e.g., Likert scales) [34]. A range of model fit statistics were examined using the following minimum thresholds defined by COSMIN: Comparative Fit Index (CFI) > 0.95; Tucker–Lewis Index (TLI) > 0.95; Root Mean Square Error of Approximation (RMSEA) < 0.06; and Standardised Root Mean Square Residual (SRMR) < 0.08 [29]. Standardised factor loadings for each item were also evaluated, with a desired minimum loading of ≥0.50 per item.
• Internal consistency: Cronbach’s α coefficient was used to confirm the internal consistency of the scale following confirmation of unidimensionality via factor analysis; a threshold of α ≥ 0.7 was used to confirm acceptable internal consistency [29].

2.6.3. Hypothesis Testing

• Known-group validity: This was undertaken to test the ability of the MobQoL-7D Dutch to discriminate between groups expected to have different outcomes. The Mann–Whitney U test was used; this compares the difference between the mean ranks of two independent groups. Mann–Whitney U is considered to be more suitable than independent t tests when comparing data from ordinal variables [35]. Our unvalidated translation of the PIADS outcome measure was used to define three pairs of “known” groups, based on the three PIADS sub-scales (Competence, Adaptability and Self-Esteem). For each sub-scale participants were grouped by whether they had experienced positive (≥0; i.e., “positive effect” group) or negative impacts (<0; i.e., “negative effect” group) associated with assistive technology use and mobility impairment. The paired groups were then compared to determine whether their answers were significantly different on each MobQoL-7D Dutch item and the overall index value, with the hypothesis that those in the “negative effect” groups would have worse outcomes comparatively. Respondents were also categorised into a further pair of “known” groups based on perceived stability of health; the two groups were defined as having “stable health” or “unstable health” based on participants’ dichotomous self-rating of health stability at baseline. Individual item scores and overall index values were assessed for significant (p < 0.05) differences between defined “known” group pairs, based on the hypothesis that those in the “unstable health” group would have worse outcomes comparatively.
• Convergent validity: This was analysed using Spearman’s Rho correlation coefficient; we examined the correlation between the MobQoL-7D Dutch and our unvalidated translations of the EQ-5D-5L and ICECAP-A. The aim was to understand whether the MobQoL-7D Dutch accurately measures specified constructs in accordance with other outcome measures. Convergent validity was determined by an item’s strongest correlation being with a hypothesised equivalent item from a comparable measure. The strength of correlations was defined as such: very weak (rs = 0.00 to 0.19), weak (rs = 0.20 to 0.39), moderate (rs = 0.40 to 0.59), strong (rs = 0.60–0.79) and very strong (rs > 0.79) [36], with moderate-to-strong correlations (>0.40) demonstrating expected convergent validity.

For ease of reporting, the individual items of the MobQoL-7D Dutch and the other outcome measures are referred to by their original English titles; for translated text, please see the MobQoL-7D Dutch in Supplementary File S2. A priori hypotheses of expected correlations were determined before analysis as follows:

• The MobQoL-7D Dutch “Accessibility” item will be most correlated with the “Mobility” item of the EQ-5D-5L (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Contribution” item will be most correlated with the “Usual activities” item of the EQ-5D-5L (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Pain/discomfort” item will be most correlated with the “Pain/discomfort” item of the EQ-5D-5L (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Independence” item will be most correlated with the “Self-care” or “Usual activities” items of the EQ-5D-5L (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Self-esteem”, “Mood/emotions” and “Anxiety” items will be most correlated with the “Anxiety/depression” item of the EQ-5D-5L (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Accessibility” item will be most correlated with the “Being independent” item of the ICECAP-A (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Contribution” item will be most correlated with the “Being independent” item or “Achievement and progress” item of the ICECAP-A (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Independence” item will be most correlated with the “Being independent” item of ICECAP-A (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Self-esteem” and “Anxiety” items will be most correlated with the “Feeling settled and secure” item of the ICECAP-A (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Mood and emotions” item will be most correlated with the “Feeling settled and secure” item or “Enjoyment and pleasure” item of the ICECAP-A (unvalidated Dutch translation).

We did not define a hypothesis for correlation between the MobQoL-7D Dutch “Pain/discomfort” item and the ICECAP-A items due to uncertainty in conceptual overlap.

3. Results

3.1. Sample and Demographics

In total 308 participants completed the survey. On average, participants were 49.6 years of age (SD = 16.0), used an average of 1.9 (SD = 1.2) mobility aids and had been a mobility aid user for an average of 7.4 years (SD = 9.7). Of the sample, 12.0% (n = 37) completed the retest evaluation at 2 weeks. Participants in the test–retest sub-group had similar key demographics to the full sample: the average age was 52.8 years (SD = 14.0), the average number of mobility aids used was 1.9 (SD = 1.2), and the average length of time as a mobility aid user was 8 years (SD = 6.8). The test–retest sub-group had better gender balance: 54.1% (n = 20) of participants were female, compared to 39.9% (n = 123) in the full sample.

Demographic details for the full sample are presented in

Table 1. Demographic details for sample.

		%	N
Gender	Female	39.9	123
	Male	59.1	182
	Other	0.3	1
	Not stated a	0.7	2
Age Category (years)	18–24	2.9	9
	25–34	18.2	56
	35–44	25.0	77
	45–54	12.7	39
	55–64	19.2	59
	65+	20.5	63
	Not stated a	1.6	5
Employment Status	Full-time	27.6	85
	Part-time	27.9	86
	Unemployed	9.7	30
	Full-time carer	3.9	12
	Retired	19.5	60
	Student	0.7	2
	Sick leave	10.1	31
	Not stated a	0.7	2
Self-rated Health Stability	Stable	70.1	216
	Unstable	28.3	87
	Not stated a	1.6	5

^a Participant chose not to provide this information.

, and details of diagnoses (using the International Classification of Diseases v10 (ICD-10) categories), primary mobility aids and frequency of mobility aid use are presented in

Table 2. Mobility aid and diagnosis details for sample.

		%	N
Frequency of Mobility Aid Use	Whole day	21.1	65
	Almost daily	49.7	153
	Weekly	17.2	53
	Monthly	8.1	25
	Not stated a	3.9	12
Primary Mobility Aid	Manual wheelchair	21.8	67
	Special support aid (insoles, support cushions, etc.)	4.5	14
	Power wheelchair (including power assist devices)	2.9	9
	Mobility scooter	14.0	43
	Bicycle (including tricycles, handbikes and special bikes)	5.2	16
	Other (toilet/shower chairs, custom seating supports, etc.)	6.8	21
	Walking aid (crutches, rollator)	11.7	36
	Car (adaptations)	11.4	35
	Home modification	1.0	3
	Not stated a	20.8	64
Primary Diagnosis: International Classification of Diseases (ICD-10) Category	I Certain infectious and parasitic diseases	0.3	1
	II Neoplasms	0.3	1
	IV Endocrine, nutritional and metabolic diseases	1.0	3
	VI Diseases of the nervous system	5.5	17
	VII Diseases of the eye and adnexa	0.7	2
	IX Diseases of the circulatory system	1.6	5
	X Diseases of the respiratory system	2.3	7
	XIII Diseases of the musculoskeletal system and connective tissue	14.6	45
	XIV Diseases of the genitourinary system	0.7	2
	XVII Congenital malformations, deformations and chromosomal abnormalities	0.7	2
	XVIII Symptoms, signs and abnormal clinical and laboratory findings, not elsewhere classified	5.5	17
	XIX Injury, poisoning and certain other consequences of external causes	4.2	13
	XXI Factors influencing health status and contact with health services	1.0	3
	Not stated a	61.7	190

^a Participant chose not to provide this information.

.

Of the sample, 28.3% (n = 87) self-reported having “unstable” health compared to 70.1% (n = 216) who self-reported having “stable” health (1.6% (n = 5) did not respond to this question). A proportion of the sample experienced negative impacts associated with assistive technology use in the three pre-defined domains (derived from our unvalidated translation of the PIADS):

• Competence: 9.4% (n = 29) experienced negative impacts on feelings of competence and efficacy relating to their assistive technology use, compared to 79.6% (n = 245) who experienced positive impacts.
• Adaptability: 13.0% (n = 40) experienced negative impacts on their willingness to try out new things and to take risks relating to their assistive technology use, compared to 76.0% (n = 234) who experienced positive impacts.
• Self-esteem: 16.6% (n = 51) experienced negative impacts on their feelings of emotional health and happiness relating to their assistive technology use, compared to 72.4% (n = 223) who experienced positive impacts.

In total 11.0% (n = 34) of the sample had missing data, which meant that sub-scale domain scores could not be calculated.

3.2. Basic Measurement Properties, Interpretability and Reliability

3.2.1. Missing Data and Floor/Ceiling Effects

See

Table 3. Summary of measurement properties, interpretability and psychometric properties of the MobQoL-7D Dutch.

		Test–Retest Reliability a	% Missing Data	Distribution of Response Choices (%)				Known-Group Validity (Significance [p] and Effect Size [r])
				Lvl 1 d	Lvl 2	Lvl 3	Lvl 4 e	Health Stability	Competence f	Adaptability f	Self- Esteem f
	MobQoL-7D Dutch Index Value	0.95 (0.91–0.98) *	2.9	12.3 b			0.0 c	p < 0.001; r = 0.50 **	p < 0.001; r = 0.34 **	p < 0.001; r = 0.29 **	p < 0.001; r = 0.26 **
MobQoL-7D Dutch Items	Accessibility	0.51 (0.10–0.92) *	0.6	56.5	37.6	5.2	0.7	p < 0.001; r = 0.35 **	p < 0.001; r = 0.24 **	p = 0.003; r = 0.18 **	p = 0.021; r = 0.14 **
	Contribution	0.74 (0.53–0.94) *	0.3	41.7	35.8	18.6	3.9	p < 0.001; r = 0.37 **	p < 0.001; r = 0.30 **	p < 0.001; r = 0.30 **	p < 0.001; r = 0.31 **
	Pain/discomfort	0.87 (0.77–0.98) *	0.3	23.1	47.9	22.8	6.2	p < 0.001; r = 0.51 **	p < 0.001; r = 0.28 **	p = 0.040; r = 0.12 **	p < 0.001; r = 0.22 **
	Independence	0.69 (0.43–0.95) *	1.3	31.3	39.1	24.3	5.3	p < 0.001; r = 0.36 **	p < 0.001; r = 0.25 **	p < 0.001; r = 0.25 **	p < 0.001; r = 0.20 **
	Self-esteem	0.78 (0.68–0.91) *	0.3	38.4	42.7	14.7	4.2	p < 0.001; r = 0.36 **	p < 0.001; r = 0.26 **	p < 0.001; r = 0.25 **	p < 0.001; r = 0.22 **
	Mood/emotions	0.82 (0.71–0.94) *	0.0	34.4	47.4	15.6	2.6	p < 0.001; r = 0.40 **	p < 0.001; r = 0.30 **	p < 0.001; r = 0.28 **	p < 0.001; r = 0.25 **
	Anxiety	0.62 (0.45–0.79) *	0.3	55.1	35.2	8.1	1.6	p < 0.001; r = 0.30 **	p < 0.001; r = 0.31 **	p < 0.001; r = 0.28 **	p < 0.001; r = 0.23 **

* Significant at p ≤ 0.001. ** = Mean rank followed the hypothesised pattern: Index value: mean rank higher for the positive effect PIADS (unvalidated Dutch translation) groups and stable health group. Individual items: mean rank is lower for the positive-effect PIADS (unvalidated Dutch translation) groups and stable-health group. ^a ICC for index values, weighted Cohen’s kappa for individual items (95% CI provided). ^b % of participants reporting the best possible health state, i.e., 1111111. ^c % of participants reporting the worst possible health state, i.e., 4444444. ^d Level 1 represents the ‘best’ response choice (i.e., no problems) for each item and is thus the ‘ceiling’—this is reversed for the index value, where lower index values equal worse health status, thus representing the ‘floor’. ^e Level 4 represents the ‘worst’ response choice (i.e., extreme problems) for each item and is thus the ‘floor’—this is reversed for the index value, where higher index values equal better health status, thus representing the ‘ceiling’. ^f Sub-scale score derived from PIADS (unvalidated Dutch translation). Cells highlighted in blue indicate potential ceiling/floor effects (i.e., >25% of response choices on the first or last response choice, respectively).

for full results. None of the MobQoL-7D Dutch items exhibited issues with missing data; the highest rate was 1.3% for the “Independence” item, which still fell well below the specified threshold of >4%; likewise, only 2.9% of participants left at least one item unanswered across all items, which again was below the specified threshold. There was no indication of floor or ceiling effects for the overall index value, with only 12.3% of participants achieving the highest possible index value (i.e., all item responses on the “best” possible response choice) and no participants achieving the lowest possible index value (i.e., all item responses on the “worst” possible response choice). At the item level, all items showed potential ceiling effects (see Figure 1), with proportions ranging from 23.1% (“Pain/discomfort”) to 56.5% (“Accessibility”). Conversely, no floor effects were observed per item, with the proportion of responses on the last response choice ranging from 0.7% (“Accessibility”) to 6.2% (“Pain/discomfort”).

Examination of only the “unstable health” sub-group (n = 87) also demonstrated a lack of floor/ceiling effects for the overall index value, with only 1.2% of participants achieving the highest possible index value and again no participants achieving the lowest (see

Table 4. Distribution of responses on each level, per MobQoL-7D Dutch item—“unstable health” sub-group (n = 87).

		Distribution of Response Choices (%)
		Lvl 1 c	Lvl 2	Lvl 3	Lvl 4 d
MobQoL-7D Dutch Index Value		1.2 a			0.0 b
MobQoL-7D Dutch Items	Accessibility	29.1	60.5	9.3	1.2
	Contribution	14.9	46.0	27.6	11.5
	Pain/discomfort	2.3	33.3	48.3	16.1
	Independence	14.0	30.2	45.4	10.5
	Self-esteem	17.2	42.5	31.0	9.2
	Mood/emotions	9.2	52.9	32.2	5.8
	Anxiety	34.5	43.7	18.4	3.5

Cells highlighted in blue indicate potential ceiling/floor effects (i.e., >15% of response choices on the first or last response choice, respectively). ^a % of participants reporting the best possible health state, i.e., 1111111. ^b % of participants reporting the worst possible health state, i.e., 4444444. ^c Level 1 represents the ‘best’ response choice (i.e., no problems) for each item and is thus the ‘ceiling’—this is reversed for the index value, where lower index values equal worse health status, thus representing the ‘floor’. ^d Level 4 represents the ‘worst’ response choice (i.e., extreme problems) for each item and is thus the ‘floor’—this is reversed for the index value, where higher index values equal better health status, thus representing the ‘ceiling’.

for full results). Per item, ceiling effects were less prevalent than for the full sample (see Figure 2), with only the “Self-Esteem” (17.2%), “Accessibility” (29.1%) and “Anxiety” (34.5%) items exceeding the ≥15% ceiling threshold. “Pain/discomfort” (16.1%) was the only item to show a potential floor effect. Overall, the results indicate increased sensitivity of the MobQoL-7D Dutch in states of worse/unstable health.

3.2.2. Test/Retest Reliability

See Table 3 for full results. For the overall index value, the ICC was 0.95 (95% CI = 0.91–0.98), demonstrating excellent test–retest reliability in this small sub-group (n = 37). Regarding the individual items, only the “Accessibility” item (κw = 0.51, 95% CI = 0.10–0.92) did not meet the criteria for adequate test–retest reliability. Weighted kappa values for the other six items ranged from κw = 0.62 to κw = 0.87, thus exceeding the stated threshold of ≥0.60.

3.3. Structural Validity

3.3.1. Exploratory Factor Analysis

The KMO measure of sampling adequacy indicated excellent suitability of the data for factor analysis (KMO = 0.90). A significant Bartlett’s test of sphericity (p < 0.001) further confirmed that EFA was appropriate. Item correlations ranged from rs = 0.38 to 0.70, indicating no obvious issues with multicolinearity and that at least one underlying latent variable was present [37].

Examination of eigenvalues and the scree plot supported a single-factor solution—the single factor had an eigenvalue of 4.19 and explained 53.4% of the variance after extraction (see

Table 5. Results from the exploratory factor analysis.

Factor	Initial Eigenvalues			Extraction Sums of Squared Loadings
	Total	% of Variance	Cumulative %	Total	% of Variance	Cumulative %
1	4.19	59.8	59.9	3.74	53.4	53.4
2	0.72	10.3	70.1
3	0.57	8.2	78.3
4	0.47	6.7	85.2
5	0.41	5.9	91.0
6	0.35	5.0	95.9
7	0.29	4.1	100

). All other eigenvalues fell below the ≥1 Kaiser criterion. The scree plot supported the presence of a single factor, with the inflexion point after the first factor (see Figure 3).

Each item contributed meaningfully to the latent variable; all seven items loaded positively on to the single factor, with standardised loadings ranging from 0.55 to 0.78 (see

Table 6. Pattern matrix: factor loadings from exploratory factor analysis.

MobQoL-7D Dutch Items	Factor Loadings *
Accessibility	0.55
Contribution	0.78
Pain/discomfort	0.73
Independence	0.76
Self-esteem	0.78
Mood/emotions	0.78
Anxiety	0.70

* Single-factor model.

).

3.3.2. Confirmatory Factor Analysis

Two confirmatory factor analyses were conducted to evaluate whether the MobQoL-7D Dutch was best represented by the single-factor structure found in the EFA or the two-factor structure found in the original MobQoL-7D validation study [16]. The single-factor model (see Figure 4) demonstrated good model fit in most tests: χ² = 51.16 (degrees of freedom ([DF] = 14; p < 0.001); CFI = 0.99, TLI = 0.98; SRMR = 0.04; RMSEA = 0.09 (90% CI = 0.07–0.12). RMSEA did not meet the stated COSMIN threshold of <0.06, likely due to the low DF. All factor loadings were significant (0.65–0.86, p < 0.001), indicating that all items loaded adequately onto the single factor.

The two-factor model (see Figure 5) also demonstrated good fit: χ² = 31.69 (df = 13; p < 0.003); CFI = 0.99, TLI = 0.99; SRMR = 0.03; RMSEA = 0.07 (90% CI = 0.04–0.10), with all loadings significant (0.66–0.88, p < 0.001). RMSEA again exceeded the stated COSMIN threshold. Correlation between the two factors was high (rs = 0.92, p < 0.001), suggesting that the two factors were not meaningfully distinct.

Overall, these results indicate that although both models show good fit, the single-factor model provides the most parsimonious representation of the MobQoL-7D Dutch; thus, the measure may be considered unidimensional. Analysis of Cronbach’s α coefficient confirmed the internal consistency of the scale (α = 0.89, 95% CI = 0.87–0.91).

3.4. Hypothesis Testing

3.4.1. Known-Group Validity

See Table 3 for full results. All items and the overall index value indicated the potential discriminative ability of the MobQoL-7D Dutch, with significant (p < 0.05) differences between known-group pairs across all tests. Effect sizes ranged from r = 0.12 (“Pain/discomfort” item, Adaptability sub-scale analysis) to r = 0.51 (“Pain/discomfort” item, health stability analysis). Mean ranks aligned with expectations: respondents with negative sub-scale scores (based on our unvalidated translation of the PIADS) and unstable health had lower mean ranks on the individual items, but higher mean ranks on the overall index value. These differences reflect the scoring structure, as higher index values represent better health status, whereas higher scores on individual items indicate greater problems in that specific domain. Analyses using the unvalidated translation of the PIADS should be viewed with caution and not assumed to be transferrable to the official Dutch translation of the PIADS.

3.4.2. Convergent Validity

See

Table 7. Convergent validity: Correlations between the MobQoL-7D Dutch items and unvalidated Dutch translations of the EQ-5D-5L and ICECAP-A items.

		EQ-5D-5L Items *					ICECAP-A Items *
		Mobility	Self-Care	Usual Activities	Pain/ Discomfort	Anxiety/ Depression	Feeling Settled and Secure	Love, Friendship and Support	Being Independent	Achievement and Progress	Enjoyment and Pleasure
MobQoL-7D Dutch items	Accessibility	0.52	0.43	0.45	0.49	0.33	0.40	0.37	0.47	0.38	0.40
	Contribution	0.63	0.47	0.59	0.55	0.46	0.49	0.36	0.56	0.54	0.50
	Pain/ Discomfort	0.66	0.46	0.65	0.79	0.32	0.45	0.39	0.52	0.61	0.42
	Independence	0.54	0.49	0.55	0.55	0.36	0.51	0.38	0.56	0.52	0.47
	Self-esteem	0.48	0.44	0.52	0.54	0.50	0.51	0.35	0.49	0.49	0.47
	Mood/ Emotions	0.49	0.50	0.54	0.54	0.56	0.54	0.41	0.46	0.51	0.53
	Anxiety	0.41	0.52	0.47	0.43	0.56	0.56	0.35	0.51	0.38	0.43

* Unvalidated Dutch translations. All correlations were found to be significant at the p < 0.01 level (2-tailed). Bold text indicates a priori hypotheses of strongest correlations for each MobQoL-7D Dutch item. Cells highlighted in blue indicate actual strongest correlations for each MobQoL-7D Dutch item.

for full results. All of the MobQoL-7D Dutch items were moderately or strongly correlated (i.e., rs > 0.40) with one or more items on our unvalidated translation of the EQ-5D-5L, and all correlations were found to be significant at the p < 0.01 level (2-tailed) and in the expected direction. A number of our a priori hypotheses were correct:

• The MobQoL-7D Dutch “Accessibility” item was most correlated with the “Mobility” item (rs = 0.52) of the EQ-5D-5L (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Pain/discomfort” item was most correlated with the “Pain/discomfort” item (rs = 0.79; these were also the most correlated items) of the EQ-5D-5L (unvalidated Dutch translation).
• The MobQoL-7D Dutch “Mood/emotions” and “Anxiety” items were most correlated with the “Anxiety/depression” item (rs = 0.56 for both) of the EQ-5D-5L (unvalidated Dutch translation).

Contrary to our hypotheses, the MobQoL-7D Dutch “Contribution” item was most correlated with the EQ-5D-5L “Mobility” item (rs = 0.63) rather than the “Usual activities” item (rs = 0.59); likewise, both the MobQoL-7D Dutch “Independence” and “Self-esteem” items were most correlated with the EQ-5D-5L “Pain/discomfort” item (rs = 0.55 and 0.54, respectively), which had not been hypothesised.

Six of MobQoL-7D Dutch items were strongly correlated with one or more items of our unvalidated translation of the ICECAP-A. All correlations were found to be significant at the p < 0.01 level (2-tailed) and in the expected direction. The strongest correlation was between the MobQoL-7D Dutch “Pain/discomfort” item and the ICECAP-A “Achievement and progress” item (rs = 0.61). All of our a priori hypotheses were correct:

• The MobQoL-7D Dutch “Accessibility”, “Contribution” and “Independence” items were all most correlated with the “Being independent” item (rs = 0.47, 0.56 and 0.56, respectively) of the ICECAP-A (unvalidated Dutch translation).
• MobQoL-7D Dutch “Self-esteem”, “Mood/emotions” and “Anxiety” items were all most correlated with the “Feeling settled and secure” item (rs = 0.51, 0.54 and 0.56, respectively) of the ICECAP-A (unvalidated Dutch translation).

As noted in the methods, analyses of convergent validity should be viewed with caution due to the use of our own unvalidated Dutch translations of the EQ-5D-5L and ICECAP-A.

4. Discussion

Overall, our preliminary analyses provide promising but not definitive evidence of the measurement properties of the MobQoL-7D Dutch. None of the MobQoL-7D Dutch items exhibited issues with missing data. Despite potential ceiling effects for all individual items, there were no obvious ceiling or floor effects for the overall index value. EFA and CFA provided good evidence for the structural validity of the MobQoL-7D Dutch as a unidimensional measure. Analysis of the “unstable health” sub-group demonstrated the potential sensitivity of the MobQoL-7D Dutch in states of worse/unstable health.

The measure exhibited excellent test–retest reliability over a 2-week period for the overall index value, but the “Accessibility” item did not meet the weighted kappa threshold for adequate reliability. Given the small sample size for the test–retest analysis, it is not clear what may have impacted the reliability of the “Accessibility” item specifically; thus, further testing is required in a larger sample to confirm these findings.

The measure demonstrated potential discriminative ability to detect differences between known groups using self-rated health stability and PIADS sub-scales to stratify the sample; due to small sub-group sample sizes, further testing is required in a larger sample to confirm these findings.

The results from the convergent validity assessment highlight how influential pain can be on various aspects of mobility, health and well-being. These results also provide a preliminary indication of good convergence between Dutch translations of the MobQoL-7D and EQ-5D-5L in relation to pain, mental well-being and mobility; and good convergence between Dutch language translations of the MobQoL-7D and ICECAP-A in relation to independence, settlement/security and mental well-being. Further analysis using the official Dutch-language versions of the EQ-5D-5L and ICECAP-A is needed to confirm these findings.

Our intention was to establish a broad and varied sample which represented the variance in users of mobility aids in the Netherlands. The sample demographics show substantial variation in ICD-10 diagnosis categories, primary mobility aids, and frequency of mobility aid use. However, we encountered issues with the completeness of the dataset; we were only able to collect data on diagnosis category, primary mobility aids and frequency of mobility aid use from a proportion of the sample. In particular, we were only able to collect diagnosis category data from less than half of our sample due to participants choosing not to report this information; we acknowledge that this potentially limits the external validity of the findings.

In the translation and analysis of the MobQoL-7D Dutch, we considered COSMIN guidance for outcome measure development and validation [18,29]. In this paper we have demonstrated a minimum of at least ‘adequate’ adherence to COSMIN study design criteria [18] for a number of factors, including general recommendations, structural validity, and internal consistency. However, due to time and funding constraints, some elements of the translation process fall short of COSMIN criteria. Specifically, cognitive interviewing and cross-cultural validation were not undertaken. These omissions may reduce confidence in the content validity of the MobQoL-7D Dutch, as cross-cultural validity and measurement invariance are critical for ensuring comparability and applicability across languages and cultures. We acknowledge that ideally cognitive interviewing should have been used to confirm the relevance, comprehensiveness and comprehensibility of the MobQoL-7D Dutch in the target population. Furthermore, we acknowledge that cross-cultural validation should have been used to determine conceptual equivalence across languages and to determine whether individual items function consistently across cultural contexts. Further research is therefore needed to assess cross-cultural validity, measurement invariance, responsiveness and further hypothesis testing before the MobQoL-7D can definitively be considered valid and reliable.

The decision to use our own unvalidated translations of the EQ-5D-5L, ICECAP-A and PIADS impacts the interpretation of the findings. The equivalence between our translations and the official versions has not been confirmed; therefore, the findings cannot be assumed to be wholly transferable to the original validated measures. As noted in the methods section, we chose this approach in order to adhere to international language standards and to ensure a level of linguistic consistency and accessibility across the outcome measures. Although the differences between our translations and the official translations may be minor, our approach may still have affected the psychometric properties of the measures. Therefore, our assessments of convergent validity and known-group validity should be considered preliminary, and further testing is now needed using the official Dutch translations of these measures to confirm these findings.

The overall sample size is reasonably large and exceeds the stated guidance for instrument validation in patient groups [27]. It should be reiterated that formal sample size calculation was not undertaken, which we acknowledge is a limitation. The sub-group sample sizes for the analysis of test–retest reliability and two of the known-group validity assessments fall below the minimum adequate sample size stated in COSMIN study design guidance for hypothesis testing [18]. Consequently, we cannot make definitive statements about the test–rest reliability and known-group validity of the MobQoL-7D Dutch; thus, the results of these analyses should be viewed as preliminary. Further hypothesis and reliability testing are therefore needed with larger samples.

Only n = 13 participants in the “unstable health” group took part in retest data collection; this sub-group was therefore considered too small to warrant reporting separate test–retest results based on self-rated health status. Future research could examine whether health stability has an impact on test–retest reliability.

The EFA and CFA in this study confirmed that the MobQoL-7D Dutch is best represented by a single-factor model. While the original two-factor model found in the validation of the English MobQoL-7D [16] also demonstrated good fit, the two latent factors were highly correlated, demonstrating that they were not meaningfully distinct in this population. This difference in factorial structure reflects the cultural and linguistic differences between the translations and the unique perspectives of the Dutch population. The single-factor model exhibits good overall fit and internal consistency according to COSMIN criteria [29], thus indicating that the MobQoL-7D Dutch is likely to be a unidimensional measure of mobility-related quality of life.

The study could have been strengthened by ensuring that the sample was demographically representative (in terms of age, sex and ethnicity) of the Dutch population; however, this was not possible due to funding and time constraints. Consequently, the sample is somewhat over-represented by males and older adults; furthermore, the sample comprised only community-dwelling adults. We also acknowledge that collecting data exclusively through an online survey may have excluded individuals with limited internet access or digital literacy and therefore impacted the representativeness of the sample.

Inclusion of a functional measure in the data collection would have been beneficial in order to test the ability of the MobQoL-7D Dutch to discriminate between defined patient groups and functional statuses. Furthermore, stratifying the sample by types of mobility aids and conditions would have provided more nuanced analyses; however, the sample size did not allow for this. These are elements we intend to explore in future aspects of this research programme through real-world application of the measure.

5. Conclusions

In conclusion, the MobQoL-7D Dutch shows promise as a new way to measure the health outcomes of individuals with impaired mobility in the Netherlands. However, further research is required to definitively confirm the validity and reliability of the MobQoL-7D Dutch. In particular, research across different patient groups and with larger sample sizes is required. In the next steps of the wider programme of research, we aim to develop a preference-based QALY scoring system for the MobQoL-7D Dutch which reflects Dutch societal preferences and to apply the MobQoL-7D Dutch in real-world clinical practice in order to further examine the usefulness of the tool.