Article Text
PDF
Abstract
Objective There is a need for a widely accepted comprehensive disease activity measure for use in daily practice in patients with psoriatic arthritis (PsA). For this reason, the 3-item Visual Analogue Scale (3VAS) and 4-item Visual Analogue Scale (4VAS) were developed. This study aimed to test construct validity and responsiveness of the 3VAS and 4VAS in a population of patients with newly diagnosed PsA receiving usual care.
Methods Components of the 3VAS (physician global, patient global, patient skin) and 4VAS (physician global, patient pain, patient joint, patient skin) were scored on 0–10 VAS scales. Agreement of low disease activity (LDA) state between 3VAS/4VAS and other composite measures was tested using Venn diagrams. Construct validity and responsiveness (3-month interval) were assessed using Spearman correlation coefficients and standardised response means (SRM) with effect sizes (ES), respectively, following hypothesis generation. Both 3VAS/4VAS were also compared with several patient-reported outcome measures.
Results Data from 629 patients were used. Both 3VAS (ES=0.48, SRM 0.52) and 4VAS (ES=0.48, SRM=0.50) showed responsiveness similar to Disease Activity in PSoriatic Arthritis (DAPSA) and Disease Activity Score-28 (DAS28). Both measures had a strong correlation with DAPSA (r=0.80–0.87), Psoriatic Arthritis Disease Activity Score (PASDAS) (r=0.89) and Routine Assessment of Patient Index Data 3 (RAPID3) (r=0.84–0.92). 3VAS and 4VAS had the highest agreement with PASDAS in categorising patients to LDA at 12 months.
Conclusion This is the first study assessing the performance of the 3VAS and 4VAS in an observational cohort of patients with early PsA. Both measures have promising performance characteristics, showing strong correlations and good discrimination with existing composite measures. The 4VAS may be the preferred version with better face validity.
- psoriatic arthritis
- autoimmune diseases
- patient reported outcome measures
Data availability statement
Data are available upon reasonable request. The data that support the findings of this study are available on request from the corresponding author FRK.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Current implementation of disease activity measures for use in patients with psoriatic arthritis (PsA) in clinical practice is slow due to lack of consensus and time constraints. For this reason, the 3-item Visual Analogue Scale (3VAS) and 4-item Visual Analogue Scale (4VAS) have been developed, which have so far shown good performance in post hoc analyses of randomised controlled trials, but have not yet been tested in observational data of patients with early PsA.
WHAT THIS STUDY ADDS
The 3-item VAS and 4-item VAS have shown to be promising composite measures for use in daily clinical rheumatology practice, showing strong correlations and good discrimination with existing composite measures.
Both the 3-item VAS and 4-item VAS showed similar responsiveness to Disease Activity Score-28 (DAS28) and Disease Activity in PSoriatic Arthritis (DAPSA) and both measures had the highest agreement with Psoriatic Arthritis Disease Activity Score (PASDAS) in categorising patients to low disease activity at 12 months.
The construct validity and responsiveness characteristics of the 3VAS and 4VAS are similar; however, the 4VAS may be the preferred version with better face validity by separate measures for joint, skin and pain in addition to the physician VAS for greater clinical utility in practice.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
This is the first study to have assessed the performance of the 3-item and 4-item VAS in early PsA, showing promising performance characteristics and therefore supports further testing of the 3VAS and 4VAS as pragmatic tools for the assessment of PsA in clinical practice.
Introduction
Psoriatic arthritis (PsA) is a heterogeneous, chronic, inflammatory disease which can lead to progressive joint destruction and deterioration of functional status, a negative impact on health-related quality of life (HRQoL) and ability to work.1 2 In the last decade, the management of PsA has improved greatly due to increased awareness among dermatologists, new treatment options and evidence for the efficacy of treat-to-target strategies. However, the incorporation of treating-to-target in daily clinical practice has been slow. This is in large part due to a lack of agreement on how to measure disease activity in routine practice. Moreover, the feasibility of current measures remains a major barrier for the further implementation of disease activity measures in daily practice.3
For this reason, the 3-item Visual Analogue Scale (3VAS) and 4-item Visual Analogue Scale (4VAS) have been developed as PsA specific, multidimensional, continuous, composite measures of disease activity. These composite scores were derived by reduction of the GRAPPA Composite Score (GRACE) measure to a 3-item VAS and a 4-item VAS version. In its derivation cohort, both versions were able to detect treatment efficacy and performed well in magnitude of change and responsiveness compared with other disease activity measures.4 The 3VAS and 4VAS have thus far been tested in post hoc analyses of a randomised controlled trial (RCT) in which guselkumab was compared with placebo. Both 3VAS and 4VAS showed a strong correlation with GRACE and Psoriatic Arthritis Disease Activity Score (PASDAS) scores. Additionally they were able to discriminate between both treatment arms.5
Further validation of the 3VAS and 4VAS could lead to implementation of this disease activity measure in daily clinical rheumatology practice, thereby improving the management of patients with PsA . However, the 3VAS and 4VAS have not yet been tested in patients with PsA seen in daily clinical practice outside the ASSESS development study.6 Therefore, the aim of this study is to test the construct validity and responsiveness of the 3-item VAS and 4-item VAS in daily clinical practice in a population of patients with newly diagnosed PsA receiving usual care.
Methods
Study design
This study followed the recommendations for analysing construct validity and responsiveness specified in the COSMIN (COnsensus-based Standards for the selection of health status Measurement INstruments)–OMERACT (Outcome Measures for Arthritis Clinical Trials) handbook.7
Patients and data collection
To test the performance of the 3VAS and 4VAS, data were used from the Dutch south west Early Psoriatic Arthritis (DEPAR) study consisting of patients with newly diagnosed PsA receiving usual care. Details of this study have been reported previously.8 Patients were seen by trained research nurses at fixed time points and data were collected on demographics, disease activity and patient-reported outcomes (including HRQoL and work productivity). For the current study, we used data from baseline, 3 months and 12 months and included patients with complete data on all 3VAS and 4VAS components at baseline.
3-item and 4-item VAS
The 3VAS and 4VAS were developed by reducing the GRACE composite measure. The GRACE index consists of eight domains including tender and swollen joint counts, patient global assessment, skin and joint VAS scores, psoriasis (Psoriasis Area and Severity Index(PASI)) and PsA quality of life.4 9 This was reduced to the physician global VAS, patient global VAS and patient skin VAS for the 3VAS and a physician global VAS, patient pain VAS, patient joint VAS and patient skin VAS for the 4VAS.6 All VAS scores range from 0 to 10. The physician global VAS is based on a full physical examination of all PsA features, including arthritis, dactylitis, enthesitis, psoriasis and axial inflammation. The sum score of all VAS components was divided by the relevant denominator to give a score range of 0–10 as a total score of the 3VAS and 4VAS. A higher score representing a higher disease activity.
Preliminary thresholds dividing both the 3VAS and 4VAS into different levels of disease activity have been published earlier and were also used for this study. The thresholds were defined in the ASSESS study (derivation cohort). The 3VAS is divided in very low disease activity (VLDA; ≤1.3), low disease activity (LDA; >1.3–≤2.4), moderate disease activity (MoDA; >2.4–<4.8) and high disease activity (HDA; ≥4.8). Similarly, the 4VAS is divided in VLDA (≤1.6), LDA (>1.6–≤2.8), MoDA (>2.8–<5.0) and HDA (≥5.0).10
Statistical analysis
For the construct validity, both the 3VAS and 4VAS, including the preliminary thresholds of meaning were tested against existing disease activity measures, including PASDAS, Disease Activity Score-28 (DAS28), Disease Activity in PSoriatic Arthritis (DAPSA), Routine Assessment of Patient Index Data 3 (RAPID3) and Minimal Disease Activity (MDA) using Spearman correlation coefficients. Spearman correlation coefficients were also used to test the correlation between the VAS composite measures and several patient-reported outcome measures (PROMs). The difference between mean scores of the 3VAS and 4VAS was tested against MDA groups using independent t-tests. A priori hypotheses were formulated based on earlier research from the DEPAR and ASSESS studies. The DEPAR study comparing several disease activity measures showed a similar performance of the PASDAS and GRACE composite measures. We formulated the following hypotheses: (1) A positive correlation was expected between both 3VAS and 4VAS with all aforementioned composite disease activity measures; (2) The strongest correlation of the 3VAS and 4VAS was expected with PASDAS; (3) The weakest correlation was expected with DAPSA due to the inclusion of mainly articular scores in DAPSA4 11; (4) 4VAS was expected to show a stronger correlation with measures of articular disease, such as DAPSA and DAS28 because it has a joint specific measure; (5) 4VAS was expected to show a stronger correlation with PROMs because it includes a relatively greater proportion of PROMs versus physician assessment. A correlation coefficient >0.5 was considered strong and >0.8 very strong.12 13
Responsiveness and the ability of the 3VAS and 4VAS to discriminate between disease states were assessed using the standardised response mean (SRM) and effect size (ES) between baseline and 3 months. Earlier research from the DEPAR study analysed responsiveness (SRM (SD), ES (SD)) of several composite disease activity measures such as DAS28 (0.83 (1.00), 0.88 (1.05)), DAPSA (0.71 (1.00), 0.73 (1.04)), GRACE (0.83 (1.00), 0.75 (0.90)) and PASDAS (0.95 (1.00), 1.00 (1.05)) in a 12-month interval.11 Responsiveness (SRM, ES) was also evaluated in the ASSESS study of PASDAS (0.84, 0.62), DAPSA (0.56, 0.44) and GRACE (0.67, 0.36). A priori hypotheses were again formulated based on these results: (1) PASDAS was expected to have the best responsiveness4 11; (2) Responsiveness of both 3VAS and 4VAS is expected to be similar to DAS28, based on the similarities between GRACE and DAS28.11 An SRM>0.80 was considered high.
To test the agreement between the 3-item and 4-item VAS and other composite scores, the proportion of patients who achieved various stages of disease activity was calculated and visualised using Venn diagrams.
The comparison instruments used were the validated measures DAPSA, PASDAS, DAS28 and MDA. RAPID3 was also included as comparison due to its feasibility in daily practice. Remission (REM) was categorised as RAPID3<3, DAPSA<4, PASDAS<1.9, DAS28<2.6. LDA was defined as RAPID3<6, DAPSA<14, PASDAS<3.2, DAS28<3.2. MoDA was categorised as RAPID3<12, DAPSA<28, PASDAS<5.4, DAS28<5.1. HDA was defined as RAPID3>12, DAPSA>28, PASDAS>5.4, DAS28>5.1. MDA was reached when >5 of 7 criteria were met.14–18 An agreement >80% was considered strong.
The correlation between the 3VAS and 4VAS with patient-reported outcomes, including the Short Form-36 (SF-36), Health Assessment Questionnaire (HAQ), 12-item Psoriatic Arthritis Impact of Disease (PSAID-12) and Bristol Rheumatoid Arthritis Fatigue (BRAF) was calculated. Work productivity was assessed using the EuroQol-5 Dimension (EQ-5D).
Missing values
We hypothesised missing data to be missing at random, therefore missing values were imputed through multiple imputations in R-4.1.2 (nset=20, JOMO package). Total scores of MDA, DAPSA, PASDAS, DAS28, RAPID3, SF36 Mental Component Summary Score, SF36 Physical Component Summary Score, PSAID-12, standard HAQ including support, EQ-5D and BRAF were imputed based on complete data of sex, age and VAS scores (VAS global from healthcare professional, VAS global patient, VAS psoriasis patient, VAS pain patient, VAS joints patient). Missing data of aforementioned variables were imputed for baseline, 3 months and 12 months. All further analyses were performed in STATA V.17.0. After imputation several checks were performed to ensure imputation was correct.
Results
Baseline characteristics
In total, 629/785 (80%) patients were included between July 2013 and March 2021. As seen in table 1, 51% (n=318) were male and the median (IQR) disease duration was 10.0 (3.6–32.6) months. Compared with patients with PsA seen in RCTs, disease activity was mild, with a median (IQR) swollen and tender joint count of 2 (1–4) and 3 (1–7), respectively. Thirty-six to forty-one per cent of patients had no enthesitis and 78% had no dactylitis. Median (IQR) PASI score in case of psoriasis was 2 (0–3). Patients reported a median (IQR) HAQ score incl. support of 0.8 (0.4–1.1).
Baseline characteristics at diagnosis of patient sample (n=629)
Agreement between levels of disease activity
The agreement between various stages of disease activity according to the different composite disease activity measures at baseline and 12 months is seen in table 2. At baseline, the 3VAS classifies 12% (n=75) of patients to VLDA, 18% (n=115) to LDA, 47% (n=292) to MoDA and 23% (n=147) to HDA. The 4VAS classifies 12% (n=77) of patients to VLDA, 20% (n=125) to LDA, 40% (n=252) to MoDA and 28% (n=175) to HDA. So, according to the 3VAS and 4VAS, 70% and 68% of patients have moderate to high disease activity, respectively. This group is the largest according to the RAPID3 (n=509, 81%) and smallest according to DAS28 (n=297, 47%).
Agreement between various stages of disease activity according to different composite disease activity measures at baseline and 12 months (n (%))
At 12 months, 44% (n=204) of patients are in VLDA, 22% (n=105) in LDA, 26% (n=122) in MoDA and 8% (n=35) in HDA according to the 3VAS. The 4VAS classifies 45% (n=211) of patients to VLDA, 20% (n=95) to LDA, 25% (n=114) to MoDA and 10% (n=46) to HDA. The pooled group of VLDA and LDA accounts for 66% of patients according to the 3VAS and 65% according to the 4VAS. DAS28 assigns the largest proportion of patients to this group (n=393, 84%) and RAPID3 the smallest (n=219, 47%).
The amount of patients with LDA (including VLDA) at 12 months is visualised in figure 1 for the 3VAS and 4VAS with other disease activity measures. The majority of patients with LDA according to PASDAS is also in LDA according to the 3VAS (n=270/291, 93%) and 4VAS (n=267/291, 92%). DAPSA and DAS28 classify 52 patients and 101 patients more to this category than the 3VAS, respectively. Compared with the 4VAS, DAPSA assigns 47 patients more to this category and DAS28 103 patients. Figure 2 shows the amount of patients with LDA (including VLDA) at 12 months according to the 3VAS and 4VAS compared with the amount of patients in MDA. This shows that 47% of patients have reached MDA. When comparing LDA (including VLDA) of the 3VAS and 4VAS to MDA, it is evident that approximately 2/3 of patients in LDA according to both VAS composite measures are also in MDA.
Agreement between LDA (including VLDA) according to various composite disease activity measures and 3-item VAS (A) and 4-item VAS (B) at 12 months (n=466).DAPSA, Disease Activity in PSoriatic Arthritis; DAS28, Disease Activity Score-28; LDA, low disease activity; PASDAS, Psoriatic Arthritis Disease Activity Score; VAS, Visual Analogue Scale; VLDA, very low disease activity.
Agreement between LDA (including VLDA) according to 3-item VAS (A) and 4-item VAS (B) and MDA at 12 months (n=466). LDA, low disease activity; MDA, minimal disease activity; VAS, Visual Analogue Scale; VLDA, very low disease activity.
Construct validity
3VAS correlated (Spearman) with 0.80 for DAPSA, 0.89 for PASDAS and 0.84 for RAPID3 (table 3). This indicates a strong correlation with DAPSA and very strong correlation with PASDAS and RAPID3. 4VAS shows a strong correlation with DAPSA (r=0.87), PASDAS (r=0.89) and RAPID3 (r=0.92). Both 3VAS and 4VAS show the weakest correlation with DAS28 (r=0.65 for 3VAS, r=0.66 for 4VAS).
Spearman correlation coefficients of 3-item VAS and 4-item VAS with existing disease activity measures at 12 months (n=466)
Next, the difference between mean scores of the 3VAS and 4VAS was tested against MDA groups using independent t-tests (table 4). Median (IQR) scores of the 3VAS were 1.1 (0.7–1.8) in the group attaining MDA and 3.6 (2.6–4.9) in the group not in MDA. For the 4VAS, these were 1.3 (0.8–1.8) in the MDA group and 4.2 (2.9–5.3) in the group not in MDA. A significant difference (p<0.001) was seen between both groups for both VAS composite measures. Regarding the correlation between the VAS composite measures and PROMs, a stronger correlation, both negatively and positively, was seen between all PROMs and the 4VAS compared 3VAS, except for the SF-36 MCS (table 5). This aligned with our set hypotheses (all further a priori hypothesised associations were found as predicted).
Median (IQR) scores of 3-item VAS and 4-item VAS per group of MDA at baseline (n=629)
Spearman correlation coefficients of 3-item VAS and 4-item VAS with several patient-reported outcome measures at 12 months (n=466)
Responsiveness
The responsiveness of existing composite disease activity measures and the 3VAS and 4VAS, including its subcomponents, is shown in table 6. The mean (SD) change of the 3VAS between baseline and 3 months was 0.86 (1.66), mean ES was 0.48 and SRM was 0.50. The 4VAS showed a mean (SD) change of 0.85 (1.69), mean ES of 0.48 and SRM of 0.52. PASDAS has the highest mean ES and SRM of 0.65 and 0.69, respectively. RAPID3 shows the lowest scores (mean ES 0.37, SRM 0.41). The mean ES and SRM of both VAS composite measures are comparable to those of DAPSA and DAS28. Aforementioned a priori hypotheses were confirmed. All SRM values were <0.80.
Responsiveness of 3-item VAS and 4-item VAS including each component and other disease activity measures between baseline and 3 months (n=508)
Last, the correlation between different levels of disease activity according to the 3VAS and 4VAS and several PROMs was evaluated at 12 months. Results are shown in table 7. In both the 3VAS and 4VAS, a worsening of PROMs was seen when disease activity increased.
Mean/median scores of several patient-reported outcome measures per 3VAS/4VAS category of disease activity at 12 months (n=466)
Discussion
This study assessed the performance of the 3-item and 4-item VAS in an observational cohort of patients with early PsA. Both these measures have shown to be promising continuous composite measures for use in daily clinical rheumatology practice, showing strong correlations and good discrimination compared with existing composite measures. The 3-item VAS and 4-item VAS showed similar responsiveness to DAS28 and DAPSA and both measures had the highest agreement with PASDAS in categorising patients to LDA at 12 months.
Treat-to-target is a valuable and proven strategy in the treatment of PsA. However, there is no consensus on which disease activity measure to use in the treatment of PsA in daily clinical practice. Currently, MDA and PASDAS are the composite measures best reflecting patient disease activity.11 However, arguably both measures are more time consuming to use, especially PASDAS. So far, only one centre worldwide was able to use the PASDAS in clinical practice.19 MDA is a dichotomous measure (a disease state that is either achieved or not), which makes it useful as a treatment target, but not for tracking disease activity over time. Moreover, MDA may be considered to be too stringent. This highlights the need for a widely accepted comprehensive disease activity measure in PsA for use in daily practice. The 3VAS and 4VAS have been tested in an RCT in which guselkumab was compared with placebo in patients with active PsA who failed on TNFα inhibitors. Data from 285 patients showed strong correlations between both VAS measures and GRACE (r=0.83–0.92) and PASDAS (r=0.72–0.85) in the intervention arm at each visit and both measures were able to discriminate between the treatment and placebo arms of the trial.5 20
Our study showed similarly strong correlations of the 3VAS/4VAS with existing composite measures, especially PASDAS, and responsiveness most comparable to DAS28 and DAPSA. DAS28 is generally not regarded as having face validity in PsA as the 28 joint count is inadequate to capture articular disease. The data presented herein showing misclassification of LDA by the DAS28 instrument (compared with the 3-item and 4-item VAS/PASDAS and DAPSA) support this view. Interestingly, both 3VAS and 4VAS showed better performance than RAPID3, which is considered to be one of the best validated patient-reported composite measures in patients with rheumatoid arthritis.21 Moreover, when comparing classification to LDA states of the 3VAS and 4VAS to being in MDA (figure 2), it is evident that MDA is more difficult to attain than LDA, according to both VAS composite measures.
The 3VAS and 4VAS have been tested in clinical trial datasets. Estimates for minimal important difference, minimal detectable change and thresholds of disease activity in a clinical trial population have been made in a pooled analysis in the DISCOVER and COSMOS studies.22 Numeric Rating Versions (NRS) of the VAS scores have been tested in a multicentre observational study, showing good correlation with HAQ and joint counts and very strong correlation with impact of disease. They have also been tested in the Upadacitinib SELECT clinical trial data, showing strong correlation with clinical and PROMs and ability to discriminate between placebo and treatment arms.23 24
Both VAS scores have shown similar performance characteristics and both balance clinician and patient perspectives in a single continuous measure, so which should be preferred for clinical practice? Work is underway to address some unanswered questions, including feasibility in clinical practice, instructions for performing the physician VAS and understanding the effect of achieving VAS LDA states on the inhibition of radiographic damage. Pending these data, and assuming these future data do not identify significant differences in performance characteristics, we suggest the 4VAS may be the preferred version. The 4VAS has separate measures for joint, skin and pain, in addition to the physician VAS. In contrast, the 3VAS comprises patient and physician global and pain, so does not provide domain specific information for use in clinical practice and therefore may be considered to have less face validity. Such differences between 3VAS and 4VAS are marginal but may help when selecting which to take forward for further testing and clinical use.
Strengths of this study include the observational study design and large number of patients. Patients enrolled in the DEPAR cohort have overall low disease activity and received usual care. This makes the DEPAR sample size representative of patients with PsA seen in daily clinical practice but can potentially limit generalisability to more active disease states. Additionally, VAS scores were not reported with the 3VAS and 4VAS composite measures in mind. We therefore believe that prospective research in real-world patients with PsA will be of added value. Different thresholds of meaning for the VAS composite measures have been estimated in the ASSESS study and RCT datasets, therefore recommended areas to focus on include further refinement of thresholds of meaning, testing feasibility of the 3VAS and 4VAS in daily practice and testing longitudinal construct validity. Additionally, it would be interesting to further decide on a preference for either the 3VAS or 4VAS in daily practice.
To conclude, both the 3-item VAS and 4-item VAS thus far have promising performance characteristics, showing strong correlations and good discrimination with existing composite measures. The data presented herein support further testing of the feasibility and performance characteristics of the 3-item VAS and 4-item VAS as pragmatic tools for the assessment of PsA in clinical practice.
Data availability statement
Data are available upon reasonable request. The data that support the findings of this study are available on request from the corresponding author FRK.
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by the local medical research ethics committee of the University Medical Center Rotterdam (MEC-2012-549) and written informed consent was obtained for all study participants according to the Declaration of Helsinki. Participants gave informed consent to participate in the study before taking part.
References
Footnotes
Contributors FRK performed the statistical analysis and drafted the manuscript. MRK, WT and MW contributed to the analysis. All authors contributed to the design and revision of the manuscript, read and approved the final manuscript. MV is the guarantor.
Funding Research support for this study was funded by the Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA) Pilot Research Grant. The organisation had no role in the study design, collection of data, analysis or interpretation of data, nor on the preparation or approval of the manuscript and the decision to submit the manuscript for publication.
Competing interests WT has received research funding, consulting or speaker fees from AbbVie, Amgen, Eli Lilly, Janssen, GSK, MSD, Novartis, Ono Pharma, Pfizer and UCB.
Provenance and peer review Not commissioned; externally peer reviewed.