Psychometric properties of the EQ-5D-5L in patients with knee or hip osteoarthritis : confirmatory factor analysis and item response theory
Kiadaliri, Ali LU
(2024)
2024 OARSI World Congress on Osteoarthritis
In Osteoarthritis and Cartilage
32(Suppl 1).
p.196-197
- Abstract
- Purpose (the aim of the study): There is limited evidence on psychometric properties of EQ-5D-5L among people with osteoarthritis (OA) undertaking non-surgical treatment. The aim of the current study was to evaluate the psychometric properties of the EQ-5D-5L among individuals with knee or hip OA participating in a digitally delivered education and exercise therapy.
Methods: Secondary analysis of data obtained from individuals aged 40+ years participating in the digital treatment during the year 2021 with the responses to EQ-5D-5L at enrolment. Confirmatory factor analysis (CFA) with the weighted least square mean and variance adjusted estimator was used to assess dimensionality. Local independency assumption was assessed using... (More) - Purpose (the aim of the study): There is limited evidence on psychometric properties of EQ-5D-5L among people with osteoarthritis (OA) undertaking non-surgical treatment. The aim of the current study was to evaluate the psychometric properties of the EQ-5D-5L among individuals with knee or hip OA participating in a digitally delivered education and exercise therapy.
Methods: Secondary analysis of data obtained from individuals aged 40+ years participating in the digital treatment during the year 2021 with the responses to EQ-5D-5L at enrolment. Confirmatory factor analysis (CFA) with the weighted least square mean and variance adjusted estimator was used to assess dimensionality. Local independency assumption was assessed using residual correlations between item pairs with r>0.2 as violation of the assumption. Monotonicity was evaluated using Mokken analysis. Item response theory was implemented by generalized partial credit model. The item parameters (discrimination and difficulty) were estimated using the maximum marginal likelihood estimation. Targeting was assessed by inspecting the item-person map (the Wright map). Differential item functioning (DIF) by sex, age (≤65 years vs. >65 years) and OA site (knee vs. hip) were assessed and items with a McFadden pseudo-R2 change of ≥0.02 were flagged for potential DIF. All analyses were conducted in RStudio version 2022.02.0.
Results: A total of 16,200 individuals with mean (standard deviation) age 64.3 (9.1), 74.0% females and 59.6% with knee OA were included. The included subjects reported 609 of the 3,125 possible health profiles of the EQ-5D-5L with slight or moderate problems with pain/discomfort dimension and no problems on the other dimensions (i.e. profiles “11121” and “11131”) being the first and second most common profiles (9.4% and 4.4% of respondents, respectively). CFA confirmed the unidimensionality of the EQ-5D-5L with factor loading ranged from 0.49 for anxiety/depression to 0.85 for usual activity dimensions, and an average variance extracted of 0.53 suggesting adequate convergent validity. Local independence and monotonicity assumptions were met. The fit statistics suggested an adequate fit of IRT model. Among five dimensions, pain/discomfort was the most affected (least difficult to endorse) while self-care was the least affected (most difficult to endorse). No disordered response thresholds were identified. EQ-5D-5L showed adequate reliability (>0.7) over a wide range of the underlying trait (76.4% of total information were within -1 to 4 standard deviation of average level of poor HRQoL) with an overall reliability of 0.80 (Figure 1). Usual activity and anxiety/depression were the most and least informative items. Inspecting the item-person map (Figure 2) suggested that while EQ-5D-5L covered the full range of HRQoL levels in the sample, it targeted, on average, a poorer HRQoL than experienced by the participants in the study. No DIF (uniform or nonuniform) was detected.
Conclusions: The EQ-5D-5L demonstrated acceptable psychometric properties among subjects with hip or knee OA participating in a digital education and exercise therapy. These results support the use of EQ-5D-5L among people with less severe knee or hip OA. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/8023c4e9-713e-400c-969f-8db5f55b1a15
- author
- Kiadaliri, Ali
LU
- organization
- publishing date
- 2024-04-01
- type
- Contribution to journal
- publication status
- published
- subject
- in
- Osteoarthritis and Cartilage
- volume
- 32
- issue
- Suppl 1
- pages
- 2 pages
- publisher
- Elsevier
- conference name
- 2024 OARSI World Congress on Osteoarthritis
- conference location
- Vienna, Austria
- conference dates
- 2024-04-18 - 2024-05-21
- ISSN
- 1063-4584
- DOI
- 10.1016/j.joca.2024.02.280
- language
- English
- LU publication?
- yes
- additional info
- This is a PROCEEDINGS ARTICLE from OARSI 2024 Conference in Vienna.
- id
- 8023c4e9-713e-400c-969f-8db5f55b1a15
- date added to LUP
- 2024-05-16 13:45:00
- date last changed
- 2024-05-16 14:06:51
@misc{8023c4e9-713e-400c-969f-8db5f55b1a15,
abstract = {{Purpose (the aim of the study): There is limited evidence on psychometric properties of EQ-5D-5L among people with osteoarthritis (OA) undertaking non-surgical treatment. The aim of the current study was to evaluate the psychometric properties of the EQ-5D-5L among individuals with knee or hip OA participating in a digitally delivered education and exercise therapy.<br/><br/>Methods: Secondary analysis of data obtained from individuals aged 40+ years participating in the digital treatment during the year 2021 with the responses to EQ-5D-5L at enrolment. Confirmatory factor analysis (CFA) with the weighted least square mean and variance adjusted estimator was used to assess dimensionality. Local independency assumption was assessed using residual correlations between item pairs with r>0.2 as violation of the assumption. Monotonicity was evaluated using Mokken analysis. Item response theory was implemented by generalized partial credit model. The item parameters (discrimination and difficulty) were estimated using the maximum marginal likelihood estimation. Targeting was assessed by inspecting the item-person map (the Wright map). Differential item functioning (DIF) by sex, age (≤65 years vs. >65 years) and OA site (knee vs. hip) were assessed and items with a McFadden pseudo-R2 change of ≥0.02 were flagged for potential DIF. All analyses were conducted in RStudio version 2022.02.0.<br/><br/>Results: A total of 16,200 individuals with mean (standard deviation) age 64.3 (9.1), 74.0% females and 59.6% with knee OA were included. The included subjects reported 609 of the 3,125 possible health profiles of the EQ-5D-5L with slight or moderate problems with pain/discomfort dimension and no problems on the other dimensions (i.e. profiles “11121” and “11131”) being the first and second most common profiles (9.4% and 4.4% of respondents, respectively). CFA confirmed the unidimensionality of the EQ-5D-5L with factor loading ranged from 0.49 for anxiety/depression to 0.85 for usual activity dimensions, and an average variance extracted of 0.53 suggesting adequate convergent validity. Local independence and monotonicity assumptions were met. The fit statistics suggested an adequate fit of IRT model. Among five dimensions, pain/discomfort was the most affected (least difficult to endorse) while self-care was the least affected (most difficult to endorse). No disordered response thresholds were identified. EQ-5D-5L showed adequate reliability (>0.7) over a wide range of the underlying trait (76.4% of total information were within -1 to 4 standard deviation of average level of poor HRQoL) with an overall reliability of 0.80 (Figure 1). Usual activity and anxiety/depression were the most and least informative items. Inspecting the item-person map (Figure 2) suggested that while EQ-5D-5L covered the full range of HRQoL levels in the sample, it targeted, on average, a poorer HRQoL than experienced by the participants in the study. No DIF (uniform or nonuniform) was detected.<br/><br/>Conclusions: The EQ-5D-5L demonstrated acceptable psychometric properties among subjects with hip or knee OA participating in a digital education and exercise therapy. These results support the use of EQ-5D-5L among people with less severe knee or hip OA.}},
author = {{Kiadaliri, Ali}},
issn = {{1063-4584}},
language = {{eng}},
month = {{04}},
note = {{Conference Abstract}},
number = {{Suppl 1}},
pages = {{196--197}},
publisher = {{Elsevier}},
series = {{Osteoarthritis and Cartilage}},
title = {{Psychometric properties of the EQ-5D-5L in patients with knee or hip osteoarthritis : confirmatory factor analysis and item response theory}},
url = {{http://dx.doi.org/10.1016/j.joca.2024.02.280}},
doi = {{10.1016/j.joca.2024.02.280}},
volume = {{32}},
year = {{2024}},
}