Article Text
PDF
Abstract
Objectives To explore the performance of the EULAR-initiated patient-reported Rheumatoid Arthritis Impact of Disease (RAID) questionnaire in relation to flares in disease activity, including comparison with other disease activity outcomes.
Methods Patients with rheumatoid arthritis in sustained remission were randomised to continued stable treatment or tapering in the ARCTIC REWIND project. In patients with flares within 12 months, we compared RAID (total score and components) at the flare visit with the visit prior to and the visit following flare, using Wilcoxon signed-rank test. Similar analyses were performed for patient global assessment, Disease Activity Score (DAS) and C reactive protein (CRP). The discriminative accuracies of RAID, patient global assessment, DAS and CRP with respect to disease activity flares were assessed by receiver operating characteristic (ROC) analyses based on logistic regression models. Flare was defined as a combination of DAS >1.6, a DAS increase ≥0.6 and ≥two swollen joints (of 44 examined) or could be recorded if patient and rheumatologist agreed that a clinically significant flare had occurred.
Results In total, 248 patients were included in the analyses, with 56 flares. RAID, patient global assessment, DAS and CRP all changed significantly at the visits related to flare (p<0.001). Area under the curve (95% CI) values indicated that RAID (0.88 (0.83 to 0.93)) was significantly more accurate than CRP (0.76 (0.69 to 0.84)) in discriminating flare, and less accurate than patient global assessment (0.92 (0.87 to 0.97)) and DAS (0.94 (0.90 to 0.98)). The RAID components with highest and lowest discriminative accuracies were pain (0.91 (0.86 to 0.95)) and sleep (0.69 (0.59 to 0.79)).
Conclusion Disease activity flares were associated with a significant increase in median RAID, supporting its ability to respond to flare.
Trial registration number NCT01881308.
- arthritis, rheumatoid
- patient reported outcome measures
- outcome assessment, health care
Data availability statement
Data are available on reasonable request.
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 ABOUT THIS TOPIC
Rheumatoid Arthritis Impact of Disease (RAID) is a patient-reported outcome measure with a potential to be applied in monitoring of disease activity in patients with rheumatoid arthritis.
WHAT THIS STUDY ADDS
This study is the first to explore the responsiveness and discriminative accuracy of the RAID score to disease flare.
Our results show that the RAID score was highly responsive and discriminative to flare.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
These results could inform for the use of RAID in monitoring of disease activity in regular clinical settings as well as in remote monitoring.
Introduction
Flares in rheumatoid arthritis (RA) disease activity are common and associated with poor clinical and patient-reported outcomes in both the short term and long term.1 2 Despite advances in RA management which have enabled a significant proportion of patients with RA to reach remission, up to 30% of patients in remission still experience recurrent flares.3 4 To facilitate early identification of flares in these patients, tight monitoring with patient-reported outcome measures (PROMs) has been suggested as supplement to prescheduled clinical visits and in remote monitoring.5 6 The perception of flare may differ between patients and healthcare professionals as patients tend to prioritise subjective symptoms such as pain, functional disability and fatigue over clinical disease activity outcomes (ie, swollen joints and acute phase reactants), thus exploring the psychometric properties of patient-reported outcomes with regard to disease activity flare is important.7–13
The Rheumatoid Arthritis Impact of Disease (RAID) score is a EULAR-initiated PROM developed in collaboration with patients with RA. It is a global composite measure of the impact of RA that takes into account pain, physical disability, fatigue, sleep disturbances, coping as well as physical and emotional well-being. The RAID score is available in >60 language versions.1 14–16
The sensitivity of the RAID score to improvement in disease activity in RA has been validated, but it is not known if RAID is sensitive to flares in disease activity.15 17 Presuming that RAID is sensitive to flares in disease activity, it could contribute to identification of flares in regular clinical settings as well as in remote monitoring.5
The objectives of this study were to explore how the total RAID score and its individual components respond to disease activity flares. Additionally, we assessed the discriminative ability to flare of the RAID score compared with other disease activity outcomes.
Methods
Study design and patients
The two randomised controlled ARCTIC REWIND trials aimed to assess the effect of disease-modifying antirheumatic drug (DMARD) tapering in patients with RA (18–80 years) in sustained remission (clinicalTrials.gov: NCT01881308). Eligible patients were diagnosed according to the American College of Rheumatology/EULAR 2010 classification criteria and had been in remission for at least 12 months on stable medication with Disease Activity Score (DAS) <1.6 and no swollen joints at inclusion. For the current analyses, we merged 1-year data from the two trials assessing tumour necrosis factor inhibitors (TNFi) and conventional synthetic DMARDs (csDMARDs), respectively. In the TNFi trial, patients were randomised 1:1 to stable TNFi or tapering to discontinuation of TNFi, with csDMARD comedication kept stable. Patients in the csDMARD trial were randomised 1:1 to stable or half dose treatment. All patients who were randomised and initiated therapy were included in the analyses. Patients were followed with visits every 4 months.18 If a flare was suspected by the patient between visits, a visit should be scheduled within a week. If a flare was confirmed, the full dose of the study medication was reinstated or treatment adjusted according to current recommendations. A flare was defined as a combination of DAS >1.6, an increase in DAS ≥0.6 units and minimum two swollen joints (of 44 examined). A clinically relevant flare could be recorded even though the formal criteria was not fulfilled if agreed on by patient and investigator. These flare definitions were used as the reference standard in determining the accuracy and responsiveness of RAID with regard to flare.
Patient and public involvement
Patient representatives were involved in the development of research questions and interpretation of the results.
Assessments
Clinical assessments
Composite disease activity measures included DAS based on 44 joints and erythrocyte sedimentation rate (ESR), Simplified Disease Activity Index and Clinical Disease Activity Index.19 Single component clinical assessments consisted of joint tenderness by Ritchie Articular Index, swollen joint count of 44 joints (SJC), physician global assessment of disease activity, ESR and C reactive protein (CRP).20
Patient-reported outcome measures
The seven components of the RAID score were measured on numeric rating scales (NRS) from 0 (best) to 10 (worst) and each assigned a relative weight on the total score in accordance with the calculation rules; pain (21%), functional disability (16%), fatigue (15%) and sleep, physical well-being, emotional well-being and coping (12%) (online supplemental material). A Patient Acceptable Symptom State (PASS) of ≤2 has been proposed for the RAID total score.21
Supplemental material
Patient global assessment of disease activity (PGA) was phrased “Please evaluate the disease activity of your joint disease during the last week. Considering all your symptoms, how would you describe your condition?” and measured on a visual analogue scale from 0 (good, no symptoms) to 100 (very poor) mm. Physical function was assessed by Patient-Reported Outcome Measurement Information System (PROMIS) physical function 20-item short form on a range from 20 to 100, translated to a T score where lower scores implied poorer outcome.22
Statistical analyses
Baseline characteristics were summarised using frequencies (percentages), mean (SD) or median (IQR) as appropriate. Only the patient’s first flare was included in the analyses. In participants who experienced a flare, median RAID (total score and components), PGA, DAS and CRP were calculated for the last visit prior to the first recorded flare, the flare visit and the first visit after flare. The median RAID at these visits was compared with the PASS threshold for RAID. The changes in the RAID total score and individual components, PGA, DAS, CRP, SJC, tender joints and PROMIS physical function between the last visit prior to flare and the flare visit were calculated and evaluated with the non-parametric Wilcoxon matched-pairs signed-rank test, and similar analyses were performed for the changes between the flare visit and the first visit after a flare. The proportions of patients with increased scores during flare and decreased scores at the visit after flare were calculated. The discriminative ability of RAID and its components to detect disease flares at a patient visit was assessed using the area under the receiver operating characteristic (ROC) curve (AUC) from a logistic regression model. The model was estimated using all visits of each patient up to and including the first flare, with the patient being censored after the first flare. Flare was treated as a dichotomous outcome at each patient visit and RAID and its change from the previous visit as continuous explanatory variables. For comparison, similar models were estimated to assess the discriminative abilities of RAID in relation to one of the most widely used global patient-reported indexes of disease activity (PGA), a highly responsive composite, clinical disease activity measure (DAS) and an acute phase reactant not included in DAS (CRP).19 23 24 To account for dependence within data from the same individual, the 95% CIs for the AUCs were calculated with subject-level bootstrapping techniques. Differences in AUC values between RAID, PGA, DAS and CRP were similarly bootstrapped to form 95% CIs. Missing data for individual components of the RAID score were imputed according to the formal scoring and calculation rules. No additional imputation of data was performed. All 95% CIs and p values were based on two-sided hypothesis tests, where a p value of <0.05 was considered statistically significant. All analyses were executed in Stata V.16.0 (StataCorp).
Results
Of the 259 patients randomised in the two ARCTIC REWIND trials, 248 patients initiated the treatment strategy, and were included in the current analyses. One hundred and fifty-nine (64%) were female and mean (SD) age was 56.1 (11.8) years, disease duration 6.3 (5.7) years and DAS 0.8 (0.3) at baseline. Median (IQR) RAID score at baseline was 0.6 (0.1–1.4) (table 1). Maximum variable-wise missing rate was <7% for all outcomes included in the analyses. The flare rate was 56/248 (22.6%). Of 56 flares recorded, 50 (89%) flares led to an intensification of DMARD therapy in agreement with the trial protocol. The remaining 6/56 (11%) flares were treated with short-term oral glucocorticoids, intra-articular injections and non-steroidal anti-inflammatory drugs.
Baseline characteristics of the study participants, n=248
Changes in RAID score and other disease activity outcomes
The median (IQR) RAID score for patients who experienced a flare was 0.9 (0.3–1.4) at the last visit prior to flare with 48/56 (86%) of scores below the PASS threshold of ≤2. At the flare visit, median (IQR) RAID was 2.6 (1.4–5.6) with 37/53 (70%) above the PASS threshold. At the first visit after flare, median RAID was 1.1 (0.1–3.1) and 26/42 (62%) of scores were below the PASS threshold for the patients with follow-up data available (figure 1A).
Outcome measures at last visit before flare, at flare visit and first visit after flare. (A) Rheumatoid Arthritis Impact of Disease, (B) patient global assessment visual analogue scale, (C) Disease Activity Score, (D) C reactive protein.* *As data were restricted to the first 12 months of follow-up a proportion of observations at the first visit after the flare was not included in the descriptive analyses. This only concerned the descriptive analyses as all visits after the first flare were excluded from the prediction analyses. Boxes indicate the first and third quartiles, the band inside the box marks the second quartile (median), while the whiskers indicate the highest and lowest values within 1.5× the IQR. Dots represent individual patients (outliers).
RAID, PGA, DAS and CRP increased significantly at the flare visit compared with the preceding visit, and subsequently decreased after the flare (p<0.0001) (table 2 and figure 1A, B, C and D). The proportion of patients with an increased score at the flare visit was 89% for RAID, 96% for DAS, 96% for PGA and 62% for CRP (table 2). At the first visit after a flare, the proportion with a decreased score was 81% for RAID, 91% for DAS, 80% for PGA and 61% for CRP for patients with follow-up data available (table 2).
Mean (SD) changes in RAID* score and components, PGA†, DAS‡, CRP, SJC‡, tender joints§ and PROMIS physical function from the last visit before flare to flare visit and from flare visit to first visit after flare
Changes in RAID components
There was a statistically significant increase in all RAID components from the last visit prior to flare to the flare visit, and a statistically significant decrease in all domains from the flare visit to the first visit after flare (figure 2 and table 2). The largest overall increase was observed in pain with a mean (SD) change of 3.1 (2.6), and 91% of the participants reported an increase in pain during a flare (figure 2 and table 2). 85% reported a decrease in physical well-being and functional disability worsened in 77% of patients during a flare. At the first visit after the flare, the components with the largest proportion of patients reporting improvement were pain (81%), functional disability (71%) and coping (69%).
Rheumatoid Arthritis Impact of Disease components at last visit before flare, flare visit and first visit after flare. Higher values indicate poorer outcome. Boxes indicate the first and third quartiles, the band inside the box marks the second quartile (median), while the whiskers indicate the highest and lowest values within 1.5× the IQR. Dots represent individual patients (outliers).
Discriminative abilities
The AUC values (95% CI) were 0.88 (0.83 to 0.93) for RAID, 0.92 (0.87 to 0.97) for PGA, 0.94 (0.90 to 0.98) for DAS and 0.76 (0.69 to 0.84) for CRP. The estimated ability to discriminate flare from non-flare corresponded to excellent AUC values for the RAID score (AUC >0.8), outstanding for DAS and PGA (AUC >0.9) and acceptable for CRP (AUC >0.7) (figure 3).25
ROC curves and AUCs (95% CI) for RAID, PGA, DAS and CRP for detection of flare based on all visits leading up to and including the first flare. AUC, area under the curve; CRP, C reactive protein; DAS, Disease Activity Score; PGA, patient global assessment of the disease activity; RAID, Rheumatoid Arthritis Impact of Disease; ROC, receiver operating characteristic;
The AUC value of the RAID score was 0.11 (0.021 to 0.20) higher than the AUC value of CRP (p=0.015), 0.045 (0.0074 to 0.082) lower than the AUC value of PGA (p=0.031) and 0.067 (0.006 to 0.13) lower than the AUC value of DAS (p=0.031).
The AUC (95% CI) values of the RAID components ranged from 0.91 (0.86 to 0.95) for pain to 0.69 (0.59 to 0.79) for sleep (figure 4).
Receiver operating characteristic curves and AUC (95% CI) values for the RAID components for detection of flare based on all visits leading up to and including the first flare. AUC, area under the curve; RAID, Rheumatoid Arthritis Impact of Disease.
Discussion
In this cohort of patients with RA in sustained remission the RAID score increased significantly in relation to disease activity flares and showed high discriminative accuracy between flare and non-flare. This study is, to our knowledge, the first to evaluate the performance of the RAID score in association with flare and could inform the use of RAID in research, clinical RA management and in remote monitoring of disease activity.
According to Outcome Measures in Rheumatology, a measure should be able to respond to changes of interest in the construct that it measures over time.26 In our analyses, the changes in the RAID score and its individual components from the visit prior to flare to the flare visit and at the following visit were statistically significant. The RAID values observed corresponded well with the suggested PASS for RAID as a majority were in an acceptable symptom state when in remission prior to a flare, whereas 70% of patients were no longer in an acceptable symptom state at the time of a flare. At the first visit after a flare most patients had returned to an acceptable symptom state. All seven RAID domains were responsive during a disease activity flare, illustrating the breadth of the impact of a flare captured by RAID.7 14 27 The larger increase in RAID pain during flare compared with the other components reflects that pain is a dominant symptom during flare.7 28
The estimated accuracy of the RAID score in discriminating flare from non-flare was excellent, reflected by a nearly 90% probability of observing an elevated RAID score during a flare. The RAID score performed significantly better than CRP, while the estimated accuracies of DAS and PGA were significantly higher than the RAID score despite the overlap in CI values between RAID, PGA and DAS. The high discriminative accuracy to flare of DAS was expected, as flares were partly defined by DAS. The PGA has been criticised for being an insufficient measure of disease activity as it has been associated with pain related to structural joint damage or central sensitisation, fatigue and functional decline rather than inflammation in established RA.29 However, in the current analyses PGA was highly accurate in detecting flare which suggests that the changes in PGA were related to disease activity in this population. A potential contributing factor could be that the phrasing of the PGA in the current analyses specifically addressed disease activity.24 Our results could indicate that a global, single domain PROM such as PGA is able to capture a clinically relevant flare in patients who have achieved stable remission for at least a year.
The discriminative accuracy with regard to flare varied across the RAID components with pain demonstrating the highest accuracy, almost corresponding to PGA, next to physical well-being and functional disability. On the other hand, sleep and emotional well-being were less accurate than CRP. These findings could indicate that even though the overall changes in each RAID component during flare were statistically significant, the smaller proportions of increased scores during flare of the sleep, emotional well-being and fatigue components might have influenced the performance of the total RAID score. The poorer accuracy of these domains could explain why the RAID score did not correspond as well with flare as PGA. The same tendency was observed at the first visit after flare where a smaller proportion of scores were reduced in the components sleep, emotional well-being and fatigue compared with pain, functional disability, physical well-being and coping.
Remote monitoring of disease activity with PROMs could improve healthcare access.5 6 In remote monitoring clinical measures such as DAS and CRP are not available, however, in our analyses both PGA and RAID were more accurate indicators of flare than the acute phase reactant, CRP, and as both PROMs demonstrated high discriminative accuracy they might be useful tools for detection of flares. While both PGA and RAID offer a global measure of disease activity, our analyses support the ability of RAID to provide valid and responsive additional information on important aspects of the disease beyond the total score, informing healthcare providers and guiding decisions regarding appropriate interventions.
Limitations of this study include that findings in a clinical trial setting with patients in deep remission and early detection and treatment of flares might not be generalisable to for example, patients in low disease activity. Second, PGA is included as a component in DAS and hence in the current flare definition, the possibility of circularity exists. Third, our analyses of change at the flare-related visits only accounted for group-level changes, not individual changes. Strengths of our analyses are the large, longitudinal dataset with frequent study visits, treatment according to EULAR recommendations and few patients lost to follow-up which enabled us to explore the association between flares and patient-reported outcomes.
In conclusion, this study provides support for the responsiveness and discriminative accuracy of the RAID score with regard to flares in disease activity. However, PGA discriminated flare more accurately than RAID and could be a preferred measure given that it is less time consuming to complete and compute. Both PGA and RAID might contribute to early identification of patients in need of an adjustment of medication related to disease flare, and further research is needed on the use of these PROMs in remote monitoring.
Data availability statement
Data are available on reasonable request.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and was approved by Regional Ethical Committee (reference number 2012/2285). Participants gave informed consent to participate in the study before taking part.
Acknowledgments
The authors wish to thank the patients who participated in the ARCTIC REWIND study, investigators, study nurses and medical staff for their contributions to the study. The authors also thank our patient partners Eli Ormerod and Ola-Jacob Sønsthagen for their contributions to this work. Abstracts summarising this research were delivered as an oral presentation during the EULAR European Congress of Rheumatology in June 202331 and featured in a poster presentation tour at the American Congress of Rheumatology in November 2023.32
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
SL and A-BA are joint senior authors.
Collaborators The ARCTIC REWIND study group: Hallvard Fremstad, Cristina Spada, Tor Magne Madland, Christian A Høili, Gunnstein Bakland, Åse Lexberg, Inger Johanne Widding Hansen, Inger Myrnes Hansen, Hilde Haukeland and Maud-Kristine Aga Ljoså.
Contributors All authors were involved in drafting the article or revising it critically for important intellectual content and approved the final manuscript to be submitted and agreed to be accountable for all aspects of the work. Conception and design of the study: KH, JS, NPS, EAH, SL and A-BA. Acquisition of data: NPS, EAH, SL, A-BA and ARCTIC REWIND study group. Analysis and interpretation of data: KH, NPS, JS, LBN, TU, TKK, DHS, EAH, SL and A-BA. KH is responsible for the overall content and accepts full responsibility for the work as the guarantor.
Funding This work was funded by Foundation Dam, research grants committee (grant number 2020/FO293936). The ARCTIC REWIND study was funded by the Research Council of Norway and South-Eastern Norway Regional Health Authority (project number 2013103). The REMEDY centre is funded by the Research Council of Norway (2012/2285) and the Olav Thon Foundation.
Competing interests All authors have completed the ICMJE uniform disclosure form at www.icmje.org/disclosure-of-interest/ (available on request from the corresponding author). KH, NPS, JS and LBN declare no conflict of interest. TU has received personal fees from Lilly, Galapagos, Pfizer and UCB. TKK has received personal fees from AbbVie, Amgen, BMS, Celltrion, Galapagos, Gilead, Grünenthal, Novartis, Pfizer, Sandoz and UCB. DHS has received salary support from research contracts unrelated to the current research from CorEvitas, Janssen, Moderna, and Novartis. He receives royalties from UpToDate on unrelated content. Also, he receives honorarium from the American College of Rheumatology for editorial work. EAH reports grants from The Research Council of Norway, grants from The South-Eastern Norway Regional Health Authority, during the conduct of the study. Also, he has received consulting fees from AbbVie, Boehringer Ingelheim, Elie Lilly, Gilead and Pfizer, and personal fees from Pfizer and UCB, outside the submitted work. SL has received a grant from Boehringer Ingelheim. ABA has received personal fees from AbbVie, Eli Lilly, Novartis and Pfizer.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.