Abstract
Objective. To examine prospectively self-reported flare characteristics and their longitudinal association with disease activity and patient-reported outcomes (PRO) in patients with rheumatoid arthritis (RA).
Methods. Consecutive RA patients with 28-joint count Disease Activity Score based on C-reactive protein (DAS28-CRP) < 3.2 and no swollen joints were examined at baseline, Month 6, and Month 12. Assessments included joint counts, DAS28-CRP, visual analog scale–evaluator’s global assessment (EGA), and PRO. Every third month, patients completed the Flare Assessment in Rheumatoid Arthritis and RA Flare Questionnaire, and disclosed self-management strategies. Flaring and non-flaring patients were compared and longitudinal associations between self-reported flare status (yes/no) and disease activity, PRO, and treatment escalation were explored.
Results. Among 80 patients with RA [74% females, mean (SD) age 63 (10) yrs, disease duration 11 (7) yrs, and baseline DAS28-CRP 1.9 (0.6)], 64 (80%) reported flare at least once during 12 months. Fifty-five percent of flares lasted less than 1 week. Common self-management strategies were analgesics (50%) and restricted activities (38%). Patients who reported being in flare had consistently higher disease activity measures and PRO compared to patients without flare. In a partly adjusted model, all flare domains, patient-reported swollen and tender joint counts and disease activity measures were associated with flares. In fully adjusted analyses, present flare was independently associated with pain (OR 1.85, 95% CI 1.34–2.60), patient-reported swollen joints (OR 1.18, 95% CI 1.03–1.36), and higher EGA (OR 1.15, 95% CI 1.04–1.28). Treatment escalation was associated with present flare (p ≤ 0.001).
Conclusion. In RA, self-reported flares were frequent, mainly managed by analgesics, substantiated by higher disease activity measures, independently associated with pain and patient-reported swollen joints, and related to treatment escalation.
Treatment of patients with rheumatoid arthritis (RA) aims at sustained remission or low disease activity (LDA)1. Owing to major advances in the treatment of RA in the last decades, these therapeutic objectives are realistic goals for many patients2. However, patients who have reached remission or LDA may still experience flares3,4. These fluctuations in disease activity are strongly associated with poor clinical outcomes, may lead to progression of radiographic joint damage and impaired function, and accelerate cardiovascular (CV) comorbidity5,6,7,8.
Following the European League Against Rheumatism (EULAR) overarching principles for the treatment of patients with RA, shared decision making between the patient and the rheumatologist is a cardinal feature in disease management1. Hence, it is essential to integrate the patient’s perspective into the flare definition. The concept of patient-reported flares has emerged as a major determinant in the disease trajectory with substantial effect on everyday life activities9,10,11, and has been linked to functional impairment4. It is well established that patient-reported flares differ from one patient to another regarding duration, frequency, and symptom severity3,12.
More recent international initiatives culminated in standardization of definition and measurement tools of flares13,14,15,16. Two questionnaires incorporating patient perspectives were developed in parallel; both aimed to identify flare domains that patients and health professionals considered important to be included in a measure of flare. The Outcome Measures in Rheumatology (OMERACT) RA Flare Group has developed the RA Flare Questionnaire (RA-FQ) and has defined RA flares as “episodes of increased RA disease activity accompanied by a cluster of symptoms of sufficient intensity and duration to require initiation of, change, or increase in therapy”13,16,17,18. A French group developed a self-administered instrument, the Flare Assessment in Rheumatoid Arthritis (FLARE-RA), to detect fluctuations in disease activity between rheumatology visits19. Both questionnaires have been validated in patients with RA14,20,21. However, a threshold for flare detection was not established at the initiation of the study for both questionnaires. Flare domains have been shown to correlate with other patient-reported outcomes (PRO) measuring similar features when analyzed cross-sectionally14,18,21. Previous studies have addressed the predictors of clinical flares defined by the Disease Activity Score (DAS) flare definition7,22. Yet evidence is sparse regarding the relationship between clinical disease activity measures and flare domains upon patient-defined flare status and regarding clinical predictors and potential drivers of self-reported flares.
The objectives of our study were to describe characteristics and self-management strategies of self-reported flares in anticitrullinated peptide antibody (anti-CCP)–and/or rheumatoid factor (RF)-positive RA patients; to assess associations between self-reported present flare, other PRO, and clinical disease activity; and to investigate whether self- reported present flare is associated with escalation of medical treatment.
MATERIALS AND METHODS
Study design and participants
The FLARA study (FLAre-in-RA) is a prospective 1-year observational single-center study, where consecutive patients with RA were recruited from the outpatient clinic at the Danish Hospital for Rheumatic Diseases between August 2016 and June 2017.
Patients ≥ 18 years were eligible if they fulfilled the American College of Rheumatology (ACR) 1987 or ACR/EULAR 2010 criteria for RA23,24, were RF- and/or anti-CCP–positive, had a 28-joint count DAS based on C-reactive protein (DAS28-CRP) < 3.2 and no clinically detectable swollen joints at baseline. Further requirements were stable disease-modifying antirheumatic drug (DMARD) treatment and no intraarticular glucocorticoid injections in the last 4 weeks prior to study entry.
The FLARA study was supported by the local patient research board and 1 member participated as a patient research partner. The study was approved by the regional ethics committee (The Regional Scientific Ethical Committees for Southern Denmark, S-20160027), and was conducted according to the Declaration of Helsinki 2013. Written informed consent was obtained from all participants.
Patient-reported flare definition
The flares were divided into present or past flare in relation to the time of completing the questionnaire. Present flare was assessed from the patient perspective by the anchor question: “Are you experiencing a flare of your RA at this time?” (yes/no) and this definition was considered the primary outcome of self-reported flare throughout the entire study.
Flare questionnaires and supplementary flare questions
Patients who reported to be in a flare, rated the flare severity on an 11-point numerical rating scale (NRS) and the flare duration (1–3, 4–7, 8–14, or > 14 days). Moreover, patients completed the RA-FQ consisting of 5 questions assessing pain, physical function, fatigue, stiffness, and participation over the past week on an NRS from 0 (none) to 10 (severe)14,20. A sum score across all items was calculated, ranging from 0 to 50. Further, patients indicated tender and swollen joints on a manikin sketch.
Past flares were identified by the question: “In the last three months (or at some time since the last visit): Do you think your RA has had a flare?” Response options were as follows: no; yes, once; and yes, more than once. This flare definition was applied to standardize the assessment of flare frequency during 1 year of followup but was not used as an outcome in any of the main analyses investigating associations between flare and other variables. Patients completed the FLARE-RA, consisting of 12 questions, and the total score was calculated, as the mean across all items, ranging from 0 (no flare) to 10 (maximum flare)25,26.
Patients who reported either past or present flares completed supplementary questions as proposed by the OMERACT RA Flare Group about flare self-management strategies such as using analgesics, reducing activities, avoiding activities, behavioral approaches, using glucocorticoids, and calling the rheumatology clinic for help15,27,28.
Patients completed the flare questionnaires when attending clinical visits at baseline, Month 6, and Month 12 after baseline, while at months 3 and 9, the questionnaires were either mailed electronically or sent as a hard copy, according to patients’ preferences. A text message reminder was sent to all patients.
Other PRO
At the clinical visits at baseline, Month 6, and Month 12, participants were asked to complete visual analog scales (VAS) for pain, fatigue, and patient’s global assessment (0–100). The Danish version of the Health Assessment Questionnaire was applied to assess physical function29.
Clinical and laboratory assessments
In accordance with the EULAR recommendations for a core dataset to support observational research, these were recorded at baseline: age, sex, disease duration, ongoing pharmacological therapy for RA, weight, height, and selected comorbidities (diabetes, hypertension, CV diseases, depression, osteoarthritis, cancer)30.
At baseline, and months 6 and 12 after baseline, a rheumatologist or a rheumatology nurse carried out a clinical examination for a swollen and tender count in 28 joints (SJC28 and TJC28, respectively), CRP was collected, and DAS28-CRP was calculated. Evaluator’s global assessment (EGA) was assessed by a VAS 0–100. The patients were tested for IgM-RF and anti-CCP positivity at baseline.
Escalation of medical treatment
The escalation of medical treatment was assessed at Month 6 and Month 12 and was defined as follows: initiation of or added and/or increased dosage of conventional DMARD (cDMARD), biological DMARD (bDMARD), or steroids, likewise treated as a binary variable (yes/no).
Statistical analysis
Descriptive statistics were presented as frequencies with percentages for categorical variables, and as means with SD for continuous variables. Self-management strategies in relation to flares were described. Depending on data distribution, parametric (2 sample t tests) or non-parametric (Wilcoxon rank-sum test) analyses were used to compare patients reporting present flare versus not in present flare. We used Fisher’s exact test to investigate the differences between treatment escalation and self-reported flare status.
We were dealing with repeated measurements on the same subjects. Thus mixed-effects logistic regression analyses were used, with subject as a random effect, to analyze associations at all timepoints simultaneously. The analyses were performed with present flare status as a binary dependent variable and the following exploratory variables: disease activity measures (SJC28, TJC28, CRP, EGA). Likewise, the analyses were performed between present flare status and the following PRO: flare domains from RA-FQ (pain, function, fatigue, stiffness, and participation) and patient-reported swollen joints and tender joints. For the clinical explanatory variables, 3 timepoints corresponding to clinical visits were used in the analyses, while for PRO, all 5 timepoints were used.
Our analyses followed a 2-step approach: first analyses with a partly adjusted and subsequently fully adjusted model. In the partly adjusted model, we included 1 explanatory variable at a time, while we included all explanatory variables simultaneously in the fully adjusted model to elucidate the independent associations. We considered age, sex, disease duration at baseline, and comorbidities as potential confounders and included them as covariates in both the partly and the fully adjusted models. Comorbidities were categorized into 3 categories: none, 1, and more than 1. To account for the time-varying variable, time (timepoints for clinical visits/patients’ reports) was treated as a categorical variable and was included in all models.
For all analyses, we reported adjusted OR estimates with 95% CI. A 95% CI excluding 1 was considered statistically significant (i.e., p < 0.05). All analyses were carried out using Stata version 15.0 (StataCorp.).
RESULTS
Eighty RA patients with mean (SD) disease duration of 11 (7) years were included. Demographic and baseline characteristics are shown in Table 1. Mean (SD) age was 63 (10) years, 74% were female, and baseline DAS28-CRP 1.9 (0.6). Patients had low levels of self-reported disease activity, only 18/80 (23%) patients had pain (VAS) above 30 mm. The majority of patients (66/80, 82%) were receiving cDMARD, 41/80 (51%) received 2 concomitant cDMARD while none was on triple therapy. Comorbidities were common: hypertension 23/80 (29%), osteoarthritis (OA) 21/80 (26%), depression 9/80 (11%), diabetes 8/80 (10%), other CV diseases 3/80 (4%), and malignancies 4/80 (5%).
Baseline characteristics of 80 patients included in the study.
Numbers of patients at each followup timepoint were 80 (100%) at 3 months, 79 (99%) at 6 months, and 78 (97.5%) at 9- and 12-month followups.
Flare characteristics
The patients completed 385 (96%) of the RA-FQ and 379 (95%) of the FLARE-RA questionnaires. Eighty percent (64/80) of the patients reported to have experienced a flare during the 12 months’ followup. Thirty-six percent (29/80) reported present flare and 71% (57/80) reported past flare at least once during 12 months. The number of present flares ranged between 1 (18 patients) and 4 (1 patient), with a mean (SD) of 1.55 (0.82) per patient. The majority of present flares (40%) had lasted 1–3 days, 15% had lasted 4–7 days, 18% remained for 8–14 days, and 27% lasted longer than 2 weeks when completing the questionnaires. The mean (SD) flare severity was 4.9 (3.0).
Metacarpophalangeal (MCP) joints were most frequently reported as being swollen or tender at the time of flares (Table 2).
Number of patients reporting swollen/tender joints across all visits when reporting flare.
Self-management strategies
Patients managed their flares through a wide variety of strategies, the most common being the use of analgesics (50%) and to reduce activities (38%). Avoiding activities and behavioral approaches (massage, attending physical therapy, exercise, applying warm/cold pack) were used in 23% and 19% of patients reporting a flare, respectively. Only a minority (15%) of the patients did not change their behavior (i.e., did not do anything differently) when they experienced a flare. The least common strategies were to call their rheumatologist/rheumatology nurse for help (11%) or to take glucocorticoids (3%). However, when reporting prolonged flares > 14 days at Month 3 and Month 9, four out of 5 patients contacted the outpatient clinic for help, which resulted in treatment escalation in 3 patients.
Comparison of patients in present flare versus not in present flare
Patients reporting present flare had higher clinical disease activity measures and higher levels of PRO, including both flare questionnaires, than patients currently not in flare. Table 3 shows the differences by flare status in variables collected at the clinical visits, while Table 4 presents differences in PRO collected every third month. All differences between flaring and non-flaring patients were statistically significant except for CRP, and fatigue at Month 6 and Month 9 (Table 3 and Table 4).
Characteristics of flaring versus non-flaring patients* at 6 and 12 months’ followup.
Patient-reported outcomes in flaring vs non-flaring* patients across all followup timepoints.
Change scores from baseline to clinical visit were higher for flaring patients compared to non-flaring patients at Month 6 and Month 12, respectively (Appendix 1).
Associations across all clinical visits between self-reported present flare and measures of disease activity
Present flare was longitudinally associated with higher disease activity measures: SJC28, TJC28, CRP, and EGA in the partly adjusted model as illustrated in Table 5. In the fully adjusted model, only EGA was significantly associated with present flare with adjusted OR (95% CI) of 1.15 (1.04–1.28, p < 0.01).
Mixed-effects logistic regression examining associations across all timepoints* with self-reported present flare as outcome variable.
Associations across all timepoints between self-reported present flare and PRO
Present flare was significantly associated with all RA-FQ flare domains, as well as with patient-reported tender and swollen joints, in the partly adjusted model (all p values < 0.001; Table 5). In the fully adjusted model, only pain and patient-reported swollen joints remained significantly associated with present flare [adjusted OR 1.85 (95% CI 1.34–2.60), p < 0.001; and 1.18 (95% CI 1.03–1.36), p < 0.05, respectively].
Associations with escalation of medical treatment at Month 6 and Month 12
At Month 6, 63% of flaring patients and 8% of non-flaring patients (p = 0.001), and at Month 12, 60% of flaring patients and 6% of non-flaring patients (p < 0.0001), were intensified in medical treatment (Table 6). None of the changes were induced by side effects.
Escalation from previous visit in antirheumatic treatment by flare status.
DISCUSSION
Our prospective study over 12 months in patients with RA in LDA showed that self-reported flares were frequent, mostly short-lived, and triggered a variety of self-management strategies. MCP joints were most frequently affected by flares. Higher joint counts, CRP, and EGA were associated with greater odds of present flare, suggesting that patient-reported flares are a marker of increased inflammation. Patient-reported flares were mainly driven by pain and self-reported swollen joints, and were associated with treatment escalation.
In our study, 36% of patients reported present flare and 71% reported past flare at least once over 12 months. These findings are in accordance with the results of an observational study in patients with established RA, where the frequency of self-reported flares ranged from 54 to 74% over a 6-month period3. In an online survey among 403 patients with RA, 95% reported a transient flare at least once during the previous 12 months12. Despite different anchor questions to detect flares and various periods of recall, previous reports and our study lend support to the notion that self-reported flares are common in RA patients. The majority (55%) of present flares were reported to last less than 1 week, consistent with the findings in a recent report on flares in RA31.
We observed that the small joints of the hands were the primary target of flare, which has not been reported previously, to the best of our knowledge. A previous study showed that ultrasonography of the hands in patients with RA in clinical remission detected subclinical inflammation in > 90% of patients32, and we speculate that subclinical inflammation depicted by sensitive imaging modalities might be a trigger of short-lived transient flares, especially in the small joints of the hand. This hypothesis needs to be examined in future studies. We have recently observed that patients who self-reported hand flares had increased inflammatory activity on ultrasonography as compared to the status when not in flare33.
Previous reports about self-management strategies among flaring patients concur with our own observations: that the primary means of self-management among flaring patients have been analgesics and reducing activities, followed by avoiding activities and behavioral approaches27,28. The least common strategy has been the use of steroids and asking the rheumatology clinic for help, consistent with our results12,27. Rarely asking for external help during flaring was also observed in a recent study evaluating the effect of a nurse-led flare management intervention, wherein for the majority of flares (62%) patients preferred self-management rather than clinical visit (32% of flares) or nursing advice over the phone (6% of flares)34.
We found that all the traditional disease activity measures such as joint counts, CRP, or EGA were associated with present flare in the partly adjusted model. However, EGA, which is thought to depict clinical signs of inflammation, was the only independently associated item in the fully adjusted model35.
Among PRO, all RA-FQ flare domains and patient-reported joints were associated with present flare, while the association was strongest for pain. This was expected because pain was recognized as a critical feature in defining a flare already at OMERACT 9 in 2008, when the process to develop a flare definition was initiated13. Pain was also found to be a key determinant of flare in a study of patient perspective on flare9.
At OMERACT 9, the investigator breakout groups recognized that swollen and tender joint counts were essential features to be included in an overall flare definition, while the patients’ breakout group did not necessarily deem synovitis as relevant13. Similarly, in a study that aimed to develop a tool to identify RA flare, not all patients mentioned joint swelling as an item to consider when defining a flare19. We observed, however, that among patients reporting to be currently in flare, patient-reported swollen joints were more strongly associated with present flare status than patient-reported tender joints, and in the fully adjusted model, patient-reported swollen joints remained independently associated with flare, while patient-reported tender joints did not. In a recent study, patients with predominantly tender joints had lower levels of inflammation as defined by ultrasonography than patients with predominantly swollen joints36. Our finding, that patients weighted swollen joints higher than tender joints while reporting a flare, adds to the evidence that patient-reported flares may indeed reflect a higher burden of inflammation.
It has previously been speculated that comorbid conditions may influence patients’ self-report of flare37. Depression may interfere with endogenous pain inhibition and enhance pain sensitivity, which is known to have a longterm effect on pain-related outcomes38. In OA, peripheral joint damage is thought to be one of the most important causes of pain39. Remarkably, in our sample self-reported flare was not associated with comorbid conditions.
We observed that patient-reported flare was associated with escalation of medical treatment. This is in line with the OMERACT flare definition, which aims at identifying clinically relevant, inflammatory flares13,16. Previously, the rheumatologist’s intention to change/intensify a treatment has been used as a proxy for RA flare22,40.
The FLARA study was neither intended nor designed to validate the existing flare questionnaires, but to investigate which patient-reported and disease activity measures were associated with self-reported status of being in flare. We examined associations between RA-FQ flare domains because this questionnaire aims to detect present flares, and we recognize the relevance of the items from the FLARE-RA, which seeks to record exacerbation that occurred between 2 visits to a rheumatology clinic and queries 12 flare domains within the preceding 3-month period. Potential associations between the FLARE-RA domains and recent or current flares were beyond the scope of our present study.
Our study has several strengths, including prospective data collection and very low attrition rate, resulting in minimal missing data. A methodological strength is that we used mixed-effects logistic regression analyses, which is appropriate for repeated measurements, because it accounts for within-subject correlations, because we were dealing with repeated observations in the same individuals. We did not calculate lags because the timepoints for serial assessment were specified a priori. However, we included the time-varying aspect in all the models.
Limitations include the single-center design, and the sample selection restricted to patients with RA who were anti-CCP–and/or RF-positive, conditions thought to herald a worse prognosis41 but that may compromise the generalizability of our findings. However, flare characteristics in our sample are comparable to previous reports, including multicenter studies recruiting the broad range of patients with RA. We used an anchor question to identify a flare and left this decision to the patient’s discretion. For analysis of our study, no validated threshold of the sum score to detect a flare was available for the RA-FQ. Only very recently, candidate thresholds were proposed for the RA-FLARE42. The aim of the analysis of the PRO was to elucidate which of the individual flare domains were the most important drivers to the notion of the patients that they were experiencing a flare. We did not use a clinician flare definition as a standard to patient-reported flare, leaving our results susceptible to single-source bias. However, self-reported flare was also associated with clinical disease activity measures.
In this 1-year followup study of patients with RA in LDA or remission, flares were frequent, triggered a broad range of self-management strategies, and were substantiated by increased disease activity measures. Patients weighted the effect of swollen joints higher than that of tender joints at the time of flare, and self-reported flare was related to escalation of medical treatment. Our findings add support to the notion that patient-reported flares may reflect the inflammatory burden of RA, and can guide treatment modification in practice.
Acknowledgment
We thank all patients who participated in this study, along with the patient research partner. We thank research secretary Kirsten Frøhlich and research radiographer Henning Jakobsen, and the Danish Hospital for Rheumatic Diseases for logistical support. We also thank the staff of the hospital for assistance with recruitment of participants.
APPENDIX 1. Changes from baseline in clinical and patient-reported outcomes by flare status.
Footnotes
This study was supported by the Danish Rheumatism Association, the University of Southern Denmark, the Region of Southern Denmark, and the Knud og Edith Eriksens Mindefond (Knud and Edith Eriksens Memorial Fund).
- Accepted for publication November 8, 2019.
REFERENCES
In this issue
