Construct Validity and Responsiveness of EQ-5D-3L in Patients With Rheumatic Disease

Construct Validity and Responsiveness of EQ-5D-3L in Patients With Rheumatic Disease

The aim of this study is to investigate the construct validity (convergent and known-groups) and responsiveness of EQ-5D-3L in patients with rheumatoid arthritis, polyarthritis, psoriatic arthritis, and ankylosing spondylitis. The study is based on retrospective registry data from the Swedish Rheumatology Registry (SRQ).

Background

Patient-reported outcome measures (PROMs) are used to measure how patients themselves experience their health and health related quality of life (HRQoL) (Sveriges kommuner och regioner 2023). One of the most commonly used PROMs is EQ-5D, which is a generic instrument used to measure, value and compare health across symptoms and diagnoses (Devlin and Brooks 2017). When choosing a PROM to follow-up care from the patient perspective and to assess HRQoL in patients with rheumatic disease, it is important to know that the instrument is valid and responsive. An instrument with low validity and responsiveness might not capture and describe patients' health and changes in their health accurately. The aim of the study is therefore to investigate the validity and responsiveness of EQ-5D-3L. The research questions are: 1. What is the construct validity of EQ-5D-3L in patients with rheumatic disease? 2. What is the responsiveness of EQ-5D-3L in patients with rheumatic disease?

Methodology

Study design and data collection

This study will assess the construct validity and responsiveness of EQ-5D-3L among patients with rheumatoid arthritis (RA), polyarthritis, psoriatic arthritis (PsA), and ankylosing spondylitis (AS) based on retrospective registry data. The design of the study will follow the established guidelines from COSMIN on how to assess validity and responsiveness of PROMs (Mokkink, Prinsen et al. 2019).

Construct validity refers to the degree to which an instrument measures the constructs it intends to measure (Fayers and Machin 2007). Construct validity will be assessed in two ways, convergent validity and known-groups validity. Convergent validity refers to how well the instrument under study (EQ-5D-3L and EQ VAS) correlates with other outcome measures (Fayers and Machin 2007). Known-groups validity refers to how well the instrument can find differences between groups known to differ. Responsiveness refers to the ability of the instrument to capture change over time in the construct that is measured (Mokkink, Terwee et al. 2010).

Historical registry data from the Swedish Rheumatology Register (SRQ) will be used to assess the construct validity and responsiveness of EQ-5D-3L and EQ VAS. Data on EQ-5D-3L has been collected since 2008 (Ernstsson, Janssen et al. 2020) and the study will include data from 2008 until the time of data extraction. The analyses will be conducted independently for the different patient groups.

Ethical approval has been granted for the project (2023-04394-01).

Outcome measures

EQ-5D-3L and EQ VAS (see text about intervention). DAS28, DAPSA, BASDAI, ASDAS. The Health Assessment Questionnaire Disability Index (HAQ-DI), the Bath Ankylosing Spondylitis Functional Index (BASFI) VAS questions (pain, fatigue, and global score)

Convergent validity EQ-5D-3L

Convergent validity will be assessed by testing hypotheses regarding the expected direction and magnitude of correlation between the EQ-5D-3L and the other outcome measures (see hypotheses in document Study Protocol and Statistical Analysis Plan). Constructs that are considered to be related are expected to have at least a moderate correlation and constructs that are considered to be similar are expected to have a strong correlation (Prinsen, Mokkink et al. 2018).

Known-groups validity EQ-5D-3L

In the assessment of known-groups validity, patients will be divided into groups for which there is an expected difference in HRQoL (see hypotheses in document Study Protocol and Statistical Analysis Plan). The groups represent patients with different levels of disease activity or functional ability.

Responsiveness EQ-5D-3L

Responsiveness will be assessed in two ways. One way is by assessing the relationship between individual changes in EQ-5D-3L index value and dimensions over time with changes in other outcome measures over the same time period (see hypotheses in document Study Protocol and Statistical Analysis Plan). The second way is to assess whether EQ-5D-3L can discriminate between patients who have improved over time and those that have not, based on changes in disease activity or functional ability. The relationship between changes in the EQ-5D-3L and changes in other outcome measures will be assessed by analysing the correlation between the changes in the variables between the two first measurements of each individual during the first year after diagnosis (see hypotheses in document Study Protocol and Statistical Analysis Plan).

To assess whether EQ-5D-3L can discriminate between patients who have improved and those that have not, the area under the receiver operating curve ROC curve (AUC) will be calculated. Patients will be considered to have improved based on results from the measures of disease activity (DAS28, DAPSA, ASDAS or BASDAI) and functional ability (BASFI and HAQ-DI) using criteria for what is defined as a response from previous studies (see hypotheses in document Study Protocol and Statistical Analysis Plan).

Convergent validity EQ VAS

Convergent validity of EQ VAS will be assessed with the same analyses used for EQ-5D-3L (see the section about convergent validity EQ-5D-3L and hypotheses in document Study Protocol and Statistical Analysis Plan).

Known-groups validity EQ VAS

Known-groups validity of EQ VAS will be assessed for the same groups as the EQ-5D-3L index (see section about Known-groups validity EQ-5D-3L and hypotheses in document Study Protocol and Statistical Analysis Plan).

Responsiveness EQ VAS

Responsiveness of EQ VAS will be assessed with the same analyses used for EQ-5D-3L (see the section about responsiveness EQ-5D-3L and hypotheses in document Study Protocol and Statistical Analysis Plan).

Inclusion Criteria:

• 18 years and older at the time of first included measurement
• A diagnosis of RA, polyarthritis, PsA, or AS
• Complete registration of responses in the EQ-5D-3L descriptive system or EQ VAS on at least one time point in SRQ
• For patients with RA: At least one measurement with Disease Activity Score 28 (DAS28) reported in relation to the same visit as EQ-5D-3L
• For patients with polyarthritis: At least one measurement with DAS28 reported in relation to the same visit as EQ-5D-3L
• For patients with PsA: At least one measurement with DAS28 or Disease Activity in Psoriatic Arthritis (DAPSA) reported in relation to the same visit as EQ-5D-3L
• For patients with AS: At least one measurement with the Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) reported in relation to the same visit as EQ-5D-3L

For the analyses of construct validity, the latest measurement will be used if the individual patients have multiple complete registrations with EQ-5D-3L and the other required measure. The hypotheses for responsiveness will be tested in patients with newly diagnosed disease (having the diagnosis for ≤12 months), as changes in disease activity are likely to be present in this group. For the analysis of responsiveness, the two first measurements during the first year will be used.