Article Text

Download PDFPDF

Original article
A quarter of patients time their early rheumatoid arthritis onset differently than physicians
  1. Leah Ellingwood1,
  2. Fatima Kudaeva1,
  3. Orit Schieir2,
  4. Susan J Bartlett3,4,
  5. Louis Bessette5,
  6. Gilles Boire6,
  7. Glen S Hazlewood7,
  8. Carol Hitchon8,
  9. Edward Keystone9,
  10. Diane Tin10,
  11. Carter Thorne11,
  12. Vivian P Bykerk12,13 and
  13. Janet Pope1
  14. on behalf of the CATCH Investigators
    1. 1Medicine, Division Rheumatology, Western University, London, Ontario, Canada
    2. 2McGill University Centre for Bioinformatics, Montreal, Québec, Canada
    3. 3Clinical Epidemiology, McGill University, Montreal, Québec, Canada
    4. 4Division of Rheumatology, Johns Hopkins, Baltimore, Maryland, USA
    5. 5Groupe de Recherche en Rhumatologie et Maladies Osseuses, Sainte-Foy, Québec, Canada
    6. 6Medicine, Division of Rheumatology, University of Sherbrooke, Sherbrooke, Quebec, Canada
    7. 7Medicine, Division of Rheumatology, University of Calgary, Calgary, Alberta, Canada
    8. 8Medicine, Division of Rheumatology, University of Manitoba, Winnipeg, Manitoba, Canada
    9. 9Medicine, Division of Rheumatology, University of Toronto Faculty of Medicine, Toronto, Ontario, Canada
    10. 10Medicine, Suthlake Regional Health Centre, Newmarket, Ontario, Canada
    11. 11Medicine, Division of Rheumatology, Southlake Regional Health Centre, Newmarket, Ontario, Canada
    12. 12Rheumatology, Hospital for Special Surgery, New York City, New York, USA
    13. 13Mount Sinai Hospital, Toronto, Ontario, Canada
    1. Correspondence to Dr Janet Pope; janet.pope{at}


    Objective Early rheumatoid arthritis (RA) treatment requires timely recognition. This large, multicentre study compared patient-reported vs physician-reported onset of early RA.

    Methods Patients from the Canadian Early ArThritis CoHort with early/suspected RA (persistent synovitis <1 year) completed questionnaires asking about the date of symptom onset; and rheumatologists date of onset for persistent synovitis. Groups with similar reported timing (patient and physician) versus differing timing of 30 days or more were compared.

    Results In 2683 patients, the median patient symptom duration (IQR) was 178 days (163) and physician-reported duration was 166 (138). 1940 (72%) patients had similar patient-reported and physician-reported onset (<30 days), whereas 497 (18%) reported onset 30 or more days preceding physicians, and 246 (9%) 30 or more days after physicians. Patients reporting onset preceding physicians had lower baseline Disease Activity Score based on 28 joint count, swollen joint counts and erythrocyte sedimentation rate (p<0.05). Patients reporting onset after physicians were more likely to be rheumatoid factor positive (p<0.001) and had higher anticitrullinated protein antibody titres (p<0.009). Regression showed low income, smoking, fibromyalgia, osteoarthritis and baseline non-methotrexate non-biological disease-modifying antirheumatic drug use were predictors for longer patient-reported symptoms. At 12 months, patients reporting longer symptom duration than physicians had lower rates of Simplified Disease Activity Index remission and higher physician global assessments.

    Conclusion Over one-fourth of patients reported differences of >1 month in symptom onset from their rheumatologist. Patients with longer symptom durations had less improvement at 1 year, which may be reflective of comorbid musculoskeletal conditions.

    • rheumatoid arthritis
    • early rheumatoid arthritis
    • symptom onset
    • early inflammatory arthritis
    • RF
    • ACPA
    • incident cohort

    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:

    Statistics from

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Key messages

    What is already known about this subject?

    • Early initiation of disease-modifying drugs in rheumatoid arthritis (RA) has significant prognostic benefit, as suggested by a therapeutic ‘window of opportunity’.

    • Definitions of disease onset in early rheumatoid arthritis (ERA) clinical studies are heterogeneous and sometimes not defined.

    • It is unknown whether timing of ERA onset differs between patients and rheumatologists; how patients with discordant onsets clinically differ; nor whether who defines the beginning of the window of opportunity might impact disease outcomes.

    What does this study add?

    • Compared with their rheumatologist, a quarter of patients reported discordant timing of RA onset (30 days or more).

    • Patients who had shorter duration compared with the physicians’ report were more likely to be seropositive.

    • Patients who reported longer symptom duration had lower rates of remission at 12 months compared with the agreement group.

    How might this impact on clinical practice?

    • Differences in patient-reported versus physician-reported symptom onset dates could have implications for defining the window of opportunity for initiating RA treatment and the likelihood of achieving treat-to-target outcomes.

    • Study findings demonstrate the importance of adopting standardised definitions of onset of ERA to enable cross-study comparisons.


    Early inflammatory arthritis (EIA) is a recent-onset arthritis with one or more swollen joints that may resolve spontaneously, develop into rheumatoid arthritis (RA) or another definite arthropathy or remain undifferentiated. RA has a prevalence of about 1% causing significant morbidity. Early initiation of disease-modifying antirheumatic drugs (DMARD) therapy in patients with RA has significant prognostic benefit as measured by increased likelihood of remission, DMARD-free remission, improved Disease Activity Score based on 28 joint count (DAS28) and reduced rates of radiologic joint destruction.1–5 DMARD therapy should be initiated at time of RA diagnosis.6 Recognition of the earliest clinically apparent stage of RA constitutes a significant focus of research and practice.7 8

    Often, in RA, initial arthritis serves as disease onset and the beginning of the therapeutic window of opportunity. However, early RA clinical studies definitions of disease onset and symptom duration are heterogeneous, imprecise and sometimes not defined, making it difficult to accurately assess early therapy outcomes.9

    Our aim was to determine if there are differences between patients and rheumatologists in reporting RA onset as the literature does not standardise which perspective we should take for the onset of RA and this could affect the window of opportunity. Previous literature does not compare patient-reported and physician-reported onsets; so, it is unclear how discordant onsets might impact disease outcomes. Based on the importance of early RA treatment initiation for achieving remission, we hypothesise that patients with discordant onsets might experience worse clinical outcomes. This study of patients with EIA from an incident cohort compared symptom onset timing as reported by patients and physicians and identified factors associated with differences in reported onset.


    Data source

    The Canadian Early Arthritis Cohort (CATCH) is a prospective observational cohort of patients with EIA from 17 Canadian recruitment sites. Enrolment criteria include age over 16 years; between 6 weeks and 12 months of persistent synovitis at entry; two or more swollen joints or one swollen metacarpophalangeal or proximal interphalangeal joint and one or more of the following: positive RF, positive anticitrullinated protein antibodies (ACPA), morning stiffness of at least 45 min, response to non-steroidal anti-inflammatory drugs or painful metatarsal phalangeal (MTP) squeeze test. Only patients with suspected early rheumatoid arthritis are enrolled. Most patients are enrolled at their first visit to a rheumatologist and the history is judged by the physician to be persistent synovitis within the time frame. If arthritis has been palindromic, the date when it became persistent is recorded as the date of onset.

    After signing informed consent, patients completed an initial questionnaire of baseline demographics and clinical characteristics which included questions on timing of symptom onset. Physicians or study coordinators independently asked patients about their symptom onset timing, particularly probing for persistent synovitis symptoms, especially in the joint that prompted the patient’s presentation. Follow-up visits with data collection are every 3 months for the initial year and subsequently every 6 months. Patients received standard care with therapy at the discretion of the treating rheumatologists and encouragement to treat to remission. All CATCH participants signed informed consent, and the respective ethics committees of all sites approved the project. DMARDs are usually started at the first visit or shortly thereafter.


    There were 2772 CATCH participants enrolled from January 2007 to March 2017. Participants were excluded if either patient-reported or physician-reported symptom onset date was missing, if reported onset date was after initial assessment date or if recorded onset was before age 16.


    Baseline variables included demographics of age, sex, ethnicity, income, smoking and education; biomarkers (ACPA, RF); number of comorbidities, comorbid osteoarthritis (OA), fibromyalgia. Baseline and 12-month variables included inflammatory markers (erythrocyte sedimentation rate (ESR), C reactive protein (CRP)); erosions, 28-swollen and tender joint counts, Clinical Disease Activity Index (CDAI), Simplified Disease Activity Index (SDAI), patient global score, physician global assessment and DAS28; oral or parenteral corticosteroids, and RA therapy. RA therapy at 3 months was also included. Baseline and 12-month remission rates were determined based on DAS28 ≤2.6, CDAI ≤2.8 and SDAI ≤3.3. Erosions were read by the rheumatologist or site radiologist scoring them as present or absent. The rheumatologists are asked for each patient the date of onset of symptoms of persistent synovitis (month/year) or unsure. The patients are asked to answer the question about the date when the first symptoms of inflammatory arthritis began. There was no training for patients or rheumatologists for these questions except the rheumatologists were told to choose the date of persistent synovitis onset and not palindromic rheumatism onset if that preceded persistent synovitis.


    Descriptive statistics were reported for baseline study characteristics and symptom durations as reported by physician and patient. The difference between physician-reported and patient-reported symptom onset in number of days was calculated. As there was variability in which day of the month was reported as onset date, and to decrease recall error, a cut-off of 30 days was used to differentiate agreement and disagreement timing groups.

    Participants were divided into three groups: <30 days difference in reported symptom onset timing (‘agreement group’); patient-reported symptom onset 30 or more days earlier than physician (‘patient earlier’) and patient-reported symptom onset 30 or more days after physician (‘patient after physician’). The three groups were compared for baseline characteristics known to be prognostic factors for EIA using one-way between-groups analysis of variance (ANOVA).

    As patients are often started on RA therapy after initial assessment, RA therapy at 3-month visit was compared. Twelve-month outcomes of DAS28, SDAI and CDAI remission, patient global score, physician global assessment, erosions, swollen joint count, tender joint count, erosions and RA therapy were also compared.

    Univariate regression analysis was performed to identify predictors of physician-onset and patient-onset timing comparing agreement within 30 days with longer by physician and longer by patient report. Variables with p value of 0.1 or less were included in stepwise multivariate analysis for predictors of discordance, and age and sex were forced variables. Analyses were repeated omitting patients who reported symptom duration of >5 years. P values <0.05 were considered to be statistically significant and 95% CIs were used. The analysis was performed using SPSS V.25.10


    Of the initial 2772 patients, 61 participants were excluded for missing physician-reported or patient-reported symptom onset dates; 16 for symptom onset prior to age 16; 12 for physician-reported or patient-reported onset after initial visit. Thus, 2683 were included in analyses. Median patient-reported symptom duration (IQR) was 178 days (163), physician-reported duration was 166 (138) days and median difference (IQR) was 0 (0). Ten per cent (n=281) of patients-reported symptom duration longer than 1 year; 1940 (72%) patients had similar patient and physician symptom onset (<30 days), whereas 743 (27%) had disagreement reported onsets of 30 days or more; 497 (18%) patients reported onset 30 or more days preceding physicians and 246 (9%) 30 or more days after physicians.

    Table 1 summarises descriptive statistics for all patients and the three onset groups as well as ANOVA results. The groups significantly differed in age, ethnicity, 2010 American College of Rheumatology criteria, initial ACPA titre, RF positivity, baseline swollen joint count, ESR, CRP, DAS28, baseline physician global assessment (all p<0.05).

    Table 1

    Baseline characteristics by disease onset group compared using ANOVA

    When there was longer patient-reported symptom duration, the outcomes at 1 year were in general not as good. The 12-month outcomes by onset group are shown in table 2. Disease activity differed at 1 year depending on concordance or discordance of RA onset, with patients reporting longer symptom duration experiencing a lower rate of SDAI remission, higher Health Assessment Questionnaire (HAQ and higher patient global assessment of disease activity than patients with agreement in onset (all p<0.05). DAS28 improvement at 12 months compared with baseline was less for patients reporting longer symptom duration than the agreement group; however, at 12 months, there were no significant differences in DAS28 or CDAI rates of remission. At 3 and 12 months, fewer patients reporting longer disease duration were on methotrexate combo therapy than the agreement group, and at 12 months more patients reporting longer disease duration were on no DMARD or biologic therapy.

    Table 2

    Twelve-month outcomes and medications by disease onset group compared using ANOVA

    In univariate linear regression analysis (table 3), non-Caucasian ethnicity, annual income <US$50 000, current smoking, comorbid OA, fibromyalgia, baseline lack of oral corticosteroid use, lack of methotrexate combination therapy and the use of non-methotrexate non-biologic DMARD were all significant predictors (p<0.1) for difference in onset towards longer patient-reported symptom duration. RF positivity, ACPA positivity and higher physician global assessment were all significant predictors (p<0.1) for difference in onset toward longer physician-reported symptom duration.

    Table 3

    Univariate analysis for predictors of discordance in onset comparing concordant onset with longer patient-reported and physician-reported symptom durations, respectively (days; negative towards longer patient-reported duration; bolded p<0.05)

    In multivariate linear regression analysis (table 4), low income, current smoking, comorbid fibromyalgia, OA and baseline use of non-methotrexate non-biologic DMARD were significant predictors for difference in onset towards longer patient-reported symptom duration. No variables remained significant predictors for discordance towards longer physician-reported symptom duration in multivariate analysis.

    Table 4

    Multivariate linear regression for predictors of longer patient-reported symptom duration (days; negative towards longer patient-reported duration)

    When we repeated the analyses excluding patients with a patient-reported symptom duration of 5 years, the same variables had significant differences between groups except there was no significant difference in baseline CRP or parenteral steroids. There was no significant difference in 12-month outcomes of patient global assessment, HAQ or SDAI remission. Patients reporting longer symptom duration had higher baseline DAS28 remission rates, and smaller changes in DAS28 from 0 to 12 months compared with the agreement group, but no difference in DAS28 at 12 months. Univariate regression analysis identified non-Caucasian ethnicity, baseline ESR, fibromyalgia and absence of baseline oral corticosteroid use as predictors of discordance towards longer patient-reported symptom (p<0.1). In multivariate models, non-Caucasian ethnicity, fibromyalgia and absence of baseline oral corticosteroids were significant predictors of difference in duration towards longer patient-reported symptoms (p<0.05). With a 5-year patient-reported symptom duration cut-off, no additional patients were excluded from regression analysis for difference towards longer physician-reported symptoms.


    In this study, a quarter of patients with RA had differing patient-reported versus physician-reported onset of 30 days or more. Of clinical importance, 12-month outcomes (SDAI remission, HAQ and patient global assessment of disease activity) differed between the onset groups. Patients who reported earlier symptoms compared with the rheumatologists had less remission and higher disease activity. They were younger with lower baseline DAS28 than the agreement group. Initially and at 12 months, they had lower rates of combination DMARDs, less corticosteroid usage and higher rates of no DMARD treatment started at first visit compared with the agreement group. These patients likely had pain from MSK conditions other than RA contributing to their disease activity scores. Whereas, patients who identified symptom onset after physicians were more likely to be rheumatoid factor positive, had higher baseline ACPA titres and higher initial physician global assessments than the agreement group, although not significant in multivariate analysis. Antibody-positive RA may present as smouldering or insidious disease that is challenging for patients to recognise11 and is associated longer time to DMARD initiation3 12–16; these patients can present later (beyond the timing of the ‘window of opportunity’). As a marker of poor prognosis, seropositive RA predicts higher disease activity, erosive disease, and functional disability.17–25

    A review of RA clinical trials that included disease duration showed studies use variable definitions ranging from onset of symptoms (symptoms rarely defined), time of first reported joint swelling, fulfilment of classification criteria, time of diagnosis and sometimes omitted any clear definition.9 Heterogeneity in definitions poses significant difficulties for ascertaining the ‘window of opportunity’. EULAR proposed recommendations for prospective cohort studies to define the onset ‘starting point’ used for reported disease/symptom duration.26 Previous publications described heterogeneous initial symptoms in patients with early rheumatoid arthritis, ranging from gradual, vague symptoms; transient, acute episodes (palindromic); migratory pain; to acute, severe and debilitating onset, fatigue, morning stiffness, impaired function and poor sleep.27–32

    Nearly 10% of patients timed their symptom onset at least 30 days after physicians. Prior to diagnosis, many patients with EIA may not be able to distinguish different types of arthritis, and misattribute early symptoms.15 28 33 34 Lack of standardisation of how onset of symptoms and persistent synovitis timing were determined constitutes a limitation; while patient baseline surveys asked for ‘date when first symptoms began’, and physician baseline surveys asked for ‘date of onset of symptoms’, neither the patients nor rheumatologists were trained about how to answer these questions. Another limitation is that there could be concordance or discordance between the reported onset of RA and some patients would be within the 3 months optimal window for best outcomes with treatment and others far outside the window within any of the groups we defined. The 30-day difference in timing of onset between patients and their rheumatologist was chosen arbitrarily.

    In multivariate analysis, OA, fibromyalgia, low annual income, active smoking and initial non-methotrexate, non-biologic DMARD use predicted discordance in reported onsets towards longer patient-reported symptom duration. OA can be associated with patients who reported RA onset prior to physicians or vice versa, but in the former group, the timing is longer. Comorbid OA and fibromyalgia as predictors of discordance reflect the difficulty in distinguishing between musculoskeletal symptoms of various aetiologies; concomitant OA or fibromyalgia predicts increased time from RA symptom onset to treatment.35 Low socioeconomic status is associated with a longer time to rheumatologist consultation, delay in DMARD initiation and worse disease activity.36–39

    Strengths of this study include a large number of EIA participants, multicentre design and real-world observational data and less recall bias as patients required new-onset fixed synovitis to enrol and completed questionnaires at their initial visit. Patients with high pain ratings seemed to be less accurate in their recall of initial symptom timing.40

    Some patients enrolled in the cohort already on therapy (eg, if enrolled after their initial rheumatology visit) so they could present with lower disease activity. Since CATCH inclusion criteria limited onset of physician-reported persistent fixed synovitis to less than a year, there could be an even greater difference in onset timings between physician and patient report in a clinic setting or if longer symptom duration were examined.


    Compared with their rheumatologist, a quarter of patients reported discordant timing of RA onset (30 days or more). Patients who had shorter duration compared with the physicians’ report were more likely to be seropositive, so they may be serologically and clinically different. Differences in patient-reported versus physician-reported symptom onset dates could have implications for the likelihood of achieving treat-to-target outcomes, as evidenced by 12-month outcomes of lower remission among patients who reported longer symptom duration compared with the agreement group. However, the comorbid conditions of OA and fibromyalgia associated with longer patient-reported disease duration likely influenced disease activity scores that are not related to inflammation from RA. Adopting standardised definitions of onset of early rheumatoid arthritis to enable cross-study comparisons is encouraged.



    • Contributors All authors contributed to reviewing the data, reviewing the paper and approving the paper. LE, FK, OS, JP performed the analyses. LE and JP wrote the initial draft. LB, GB, GSH, CH, EK, DT, CT, VPB, JP all collected data on their patients in CATCH.

    • Funding The CATCH study was designed and implemented by the investigators and financially supported through unrestricted research grants from: Amgen and Pfizer Canada—Founding sponsors since January 2007; UCB Canada, AbbVie Corporation and Bristol-Myers Squibb Canada since 2011; Medexus Inc. since 2013; Eli Lilly Canada since 2016; Merck Canada since 2017 and Sandoz Canada Pharmaceuticals since 2018. Previously funded by Hoffmann-LaRoche and Janssen Biotech from 2011 to 2016, and Sanofi Genzyme from 2016 to 2017.

    • Competing interests None declared.

    • Patient consent for publication Not required.

    • Ethics approval All sites had institutional review board approval. All patients signed informed consent. The study was conducted according to Declaration of Helsinki.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data availability statement Data are available on reasonable request.