Rheumatology International

, Volume 34, Issue 1, pp 85–92

Factors associated with time to diagnosis in early rheumatoid arthritis


    • Departments of Medicine and Community Health SciencesUniversity of Calgary
  • Juan Xiong
    • Rheumatology, Mount Sinai HospitalUniversity of Toronto
  • Janet E. Pope
    • Department of MedicineUniversity of Western Ontario
  • Gilles Boire
    • Département de MédecineUniversité de Sherbrooke
  • Carol Hitchon
    • Department of MedicineUniversity of Manitoba
  • Boulos Haraoui
    • Department of RheumatologyCentre Hospitalier de l’Universite de Montreal
  • J. Carter Thorne
    • Southlake Regional Health Centre
  • Diane Tin
    • Southlake Regional Health Centre
  • Edward C. Keystone
    • Department of MedicineUniversity of Toronto
  • Vivian P. Bykerk
    • Rheumatology, Mount Sinai HospitalUniversity of Toronto
    • Hospital for Special SurgeryWeill Cornell Medical College
  • Canadian early ArThritis CoHort (CATCH) Study Investigators
Original Article

DOI: 10.1007/s00296-013-2846-5

Cite this article as:
Barnabe, C., Xiong, J., Pope, J.E. et al. Rheumatol Int (2014) 34: 85. doi:10.1007/s00296-013-2846-5


Early diagnosis and treatment yield optimal outcomes in rheumatoid arthritis (RA); thus, barriers to disease recognition must be identified and addressed. We determined the impact of sociodemographic factors, medical comorbidities, family history, and disease severity at onset on the time to diagnosis in early RA. The Canadian early ArThritis CoHort study data on 1,142 early RA patients were analyzed for predictors of time to diagnosis using regression analysis. Sociodemographic factors (age, sex, income strata, education, ethnicity), measures of disease activity (joint counts, DAS28 score, acute-phase reactants, patient global evaluation, function), family history, serology, chronic musculoskeletal and mental health conditions, and obesity at diagnosis were considered. In multivariate linear regression analysis, more swollen joints (β = −0.047 per joint, 95 % CI −0.085, −0.010, p = 0.014), higher erythrocyte sedimentation rate (ESR) (β = −0.012 per 1 mm/h, 95 % CI −0.022, −0.002, p = 0.0018), and worse patient global scores (β = −0.082 per 1 unit on a visual analogue scale, 95 % CI −0.158, −0.006, p = 0.034) at baseline predicted a shorter time to diagnosis. Anti-cyclic citrullinated peptide (anti-CCP) antibody positivity (β = 0.688, 95 % CI 0.261, 1.115, p = 0.002) and low income (annual <$20,000 β = 1.185, 95 % CI 0.227, 2.143, p = 0.015; annual $20,000–50,000 β = 0.933, 95 % CI 0.069, 1.798, p = 0.034) increased time to diagnosis. In the logistic regression models, the odds of being diagnosed within 6 months of symptom onset were increased for each swollen joint present [odds ratio (OR) 1.04, 95 % CI 1.02–1.06 per joint], each 1 mm/h elevation in the ESR (OR 1.01, 95 % CI 1.00–1.02), and decreased for patients who were either rheumatoid factor or anti-CCP positive compared to both factors being negative (OR 0.68, 95 % CI 0.51–0.91). Higher disease activity results in a more rapid diagnosis for Canadian patients with early RA, but those with lower income have delays in diagnosis. Strategies to identify patients with a less severe disease presentation and in lower socioeconomic strata are needed to ensure equal opportunity for optimal management.


Rheumatoid arthritisEarly treatmentAccess to careHealth accessibility


The duration of time from symptom onset to diagnosis is a modifiable determinant of joint damage and affects the odds of entering remission and the long-term treatment outcomes of patients with rheumatoid arthritis (RA) [13]. Delays in patient presentation to primary care providers, subsequent referral for rheumatology assessment, and appointment availability with the rheumatologist will all impact early appropriate initiation of disease-modifying therapy. Factors impacting this time that can be modified must be determined to ensure optimal outcomes are achieved.

The majority of the delay has been attributed to presentation to primary care and subsequent referral to specialty care [4]. To date, evaluation of determinants of time to diagnosis and/or treatment initiation in RA has included patient demographic features such as age, sex, ethnicity, comorbidities, education and socioeconomic status, a family history of RA, clinical parameters of joint involvement, acute-phase reactants, serology results, and consultant practice features [410]. Conflicting results are apparent in these studies, attributable to the spectrum of clinical settings in which they have been performed, the selection of determinants studied, and study methodology. In particular, these previous studies have been performed in the hospital setting, through retrospective chart review or through administrative data analysis, and few studies have evaluated these factors in the setting of an observational study of usual clinical care with new onset RA. In addition, factors that make it difficult to properly assess patients for RA in the primary care setting have not been examined. For example, obesity can impact the interpretation of joint examination. The impact of mental health conditions (such as depression) on symptom description has not been considered. Our objective was to incorporate these new aspects in an analysis of predictors of time to diagnosis and broaden the scope of measures of disease activity evaluated in an observational prospective early inflammatory arthritis cohort.


Data source

The Canadian early ArThritis CoHort (CATCH) study is a prospective multicenter cohort of patients with early inflammatory arthritis [11]. Patients enrolled are over the age of 16 with ≥2 swollen joints or 1 swollen metacarpophalangeal or proximal interphalangeal joint, with a persistent symptom duration of 6 weeks to 12 months, and ≥1 of: positive rheumatoid factor (RF), anti-cyclic citrullinated peptide (anti-CCP), prolonged morning stiffness (>45 min), response to anti-inflammatories, or painful metatarsophalangeal squeeze test. At enrollment, 73.4 % of patients met the 1987 American Rheumatism Association (ARA) classification criteria for RA [12], and 85.5 % of patients met the 2010 American College of Rheumatology/European League Against Rheumatism (ACR/EULAR) classification criteria [13]. For this analysis, data on patients with RA by one of these criteria at enrollment were included. At baseline, patient demographics, history of musculoskeletal symptom onset, and patient-reported global evaluation of disease activity and function as measured by the health assessment questionnaire (HAQ) [14] are determined through a patient questionnaire verified by research assistants. RF and anti-CCP test are performed if not done previously. Patient assessments are performed at baseline and every 3 months for the first year and then every 6 months and include examination of tender and swollen joints by a rheumatologist, and measurement of acute-phase reactants. Patients are treated at the discretion of their rheumatologists who are encouraged to treat to achieve clinical remission.


The time to diagnosis is defined as the duration of time from self-reported symptom onset to the diagnosis date provided by the rheumatologist. This integrates the time it takes for a patient to seek care, a primary care provider to evaluate and refer, and the wait time to see the specialist. If symptoms were initially intermittent, the timing of initial persistent arthritis was used.

Predictors of time to diagnosis

In this analysis, we evaluated whether patient age, sex, annual income at presentation (in Canadian dollars; <$20,000 per year, $20,000–50,000, $50,000–100,000, >$100,000), highest education level completed (elementary, high school, college or trade school, or university), or ethnicity (Caucasian or minority) affected time to diagnosis. Measures of disease activity at the time of initial presentation to the cohort including tender and swollen joint counts (28 and 68/66 joints), DAS28 [15], acute-phase reactants [erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP)], patient global evaluation of disease, and HAQ score [14] were analyzed. We evaluated whether a family history of RA and baseline serology (RF, anti-CCP) affected time to diagnosis, and as well evaluated the impact of obesity (using body mass index (BMI) at the baseline visit), smoking status at baseline, self-reported history of mental health conditions (depression, generalized anxiety disorder, etc.), and other painful musculoskeletal conditions (such as osteoarthritis or fibromyalgia) on time to diagnosis.


Descriptive statistics (mean and standard deviation for continuous variables; frequency and percentage for categorical variables) were used to summarize the baseline variables included in this analysis. Linear regression was used to determine the significant predictors for time to diagnosis. Simple linear regression was fitted to each potential factor, and factors at the 10 % significance level were entered in the multivariate linear regression model where stepwise selection methodology was applied to yield the best model fit retaining the sociodemographic variables of interest. Logistic regression models were also analyzed to determine the odds ratios associated with diagnosis in the first 6 months of disease (yes/no). Statistical significance was defined as p < 0.05 (two-sided) at the multivariate stage. All statistical analyses were performed using SAS® (version 9.2).


As of April 2012, a total of 1,142 patients with RA at enrollment by ACR criteria (either 1987 or 2010) were recruited to CATCH. The mean symptom duration at the baseline cohort visit was 5.9 months (SD 3.0, range 1 day to 1 year) with a median of 5.5 months. The cohort characteristics are summarized in Table 1.
Table 1

Baseline variables assessed in univariate analysis for time to diagnosis


CATCH cohort to April 2012

Patients diagnosed in first 6 months of disease

Patients diagnosed in second 6 months of disease

p value*







842 (73.7 %)

422 (74.2 %)

420 (73.3 %)


Age, years

53.7 (14.6)

55 (15)

53 (15)


Caucasian ethnicity

952 (83.4 %)

468 (82.2 %)

484 (84.5 %)


Body mass indexa

28.6 (10.6)

28.8 (13.4)

28.4 (7.2)



5.04 (1.44)

5.3 (1.39)

4.8 (1.45)


Swollen joint count (28 joints)

7.8 (6.0)

8.6 (6.2)

6.9 (5.6)


Tender joint count (28 joints)

8.4 (6.5)

9.1 (6.6)

7.7 (6.3)


Swollen joint count (66 joints)

9.9 (7.9)

10.9 (8.2)

8.9 (7.5)


Tender joint count (68 joints)

12.7 (9.4)

13.5 (9.3)

12.0 (9.5)


HAQ score (0–3)

1.02 (0.70)

1.10 (0.72)

0.95 (0.67)


Patient global (VAS 0–10)

5.7 (2.9)

6.0 (2.9)

5.5 (2.9)


ESR, mm/h

28.2 (22.7)

31.0 (23.7)

25.3 (21.2)


CRP, mg/L

14.5 (17.9)

16.6 (19.4)

12.3 (15.9)


Rheumatoid factor positiveb

700/1,050 (67.3 %)

334 (64.6 %)

366 (70.0 %)


Anti-CCP positiveb

466/813 (58.0 %)

200 (52.1 %)

266 (63.3 %)


Either rheumatoid factor or anti-CCP positive

783/1,079 (72.6 %)

367 (68.2 %)

416 (76.9 %)


Education (highest level)

Elementary or high school: 523 (45.8 %)

266 (46.4 %)

257 (45.2 %)


College or post-secondary: 575 (50.3 %)

276 (48.2 %)

299 (52.5 %)

Annual income

<$20,000: 132 (11.6 %)

66 (11.6 %)

66 (11.6 %)


$20,000–50,000: 277 (24.3 %)

133 (23.2 %)

144 (25.3 %)

$50,000–100,000: 198 (17.3 %)

96 (16.8 %)

102 (17.9 %)

>$100,000: 94 (8.2 %)

48 (8.4 %)

46 (8.1 %)

No answer: 441 (38.6 %)

230 (40.1 %)

211 (37.1 %)

Smoking status

Never smoke: 475/1,139 (41.7 %)

250 (43.9 %)

225 (39.5 %)


Current smoker: 213/1,139 (18.7 %)

93 (16.3 %)

120 (21.1 %)

Ex-smoker: 451/1,139 (39.6 %)

227 (39.8 %)

224 (39.4 %)

Mental health condition

Depression: 123 (11.4 %)

65 (11.4 %)

58 (10.2 %)


Other: 7 (0.6 %)

3 (0.5 %)

4 (0.7 %)


25 (2.2 %)

11 (1.9 %)

14 (2.5 %)



131 (11.5 %)

63 (11.1 %)

68 (12.0 %)


Family history of RA

245 (21.6 %)

125 (22.0 %)

120 (21.2 %)


Results are given as mean (SD) or %

p value for the comparison between patients diagnosed in the first versus second 6 months of clinical disease, uncorrected for multiple comparisons

aData only available for 650 of the 1,142 patients

bAccording to local laboratory

In univariate linear regression analysis, age, ethnicity, joint counts (swollen and tender), DAS28, patient global, HAQ, ESR, CRP, education level, being a non-smoker, being either RF or anti-CCP positive, and anti-CCP status alone were all significant predictors (p < 0.10) for time to diagnosis (Table 2), whereas BMI, sex, income, being RF positive, family history of RA, and history of depression, fibromyalgia or osteoarthritis had no effect.
Table 2

Univariate analysis for predictors of time to diagnosis (per month)



Standard error

p value

Age (per year)




Ethnicity (Caucasian)




BMI (per unit increase)




Sex (female)




Swollen joint count (28 joints, per joint)




Tender joint count (28 joints, per joint)




DAS28 (per 1 unit increase)




Patient global (per 1 unit on 0–10 visual analogue scale)




HAQ (per 1 unit increase)




ESR (per 1 mm/h increase)




CRP (per 1 mg/L increase)




Rheumatoid factor (positive)




Anti-CCP (positive)




Either rheumatoid factor or anti-CCP positive




Family history of RA (none)




Absence of depression




Absence of fibromyalgia




Absence of osteoarthritis




Education level





 High school












Income level

 No answer
















Smoking status

 Never smoke








BMI body mass index, DAS28 disease activity score based on a 28 joint count, HAQ health assessment questionnaire, ESR erythrocyte sedimentation rate, CRP c-reactive protein; anti-CCP anti-cyclic citrullinated peptide antibody, RA rheumatoid arthritis

In multivariate linear regression analysis, higher swollen joint counts, higher ESR, and worse patient global scores were significant predictors of shorter time to diagnosis (Table 3). Our analysis also found that anti-CCP antibody positivity and lower income levels were associated with a longer time to diagnosis (Table 3).
Table 3

Multivariate analysis for predictors of time to diagnosis (per month)



95 % CI

p value



4.296, 7.463


Sex (female)


−0.605, 0.363


Ethnicity (Caucasian)


−0.265, 0.887


Education level



−0.960, 1.637


 High school


−0.934, 1.357




−0.749, 1.578




−0.530, 1.862


Income level

 No answer


−0.460, 1.164




0.227, 2.143




0.069, 1.798




−0.374, 1.401



 Never smoke


−0.906, 0.285




−0.724, 0.453


Swollen joint count (28 joints), per joint


−0.085, −0.010


ESR, per 1 mm/h increase


−0.022, −0.002


Patient global score, per 1 unit on 0–10 visual analogue scale


−0.158, −0.006


Anti-CCP (positive)


0.261, 1.115


ESR erythrocyte sedimentation rate, Anti-CCP anti-cyclic citrullinated peptide antibody

We also considered the determinants of time to diagnosis when anti-CCP antibody testing was not performed, as availability of the test is restricted in particular regions of Canada and is not typically ordered by primary care physicians, and therefore would not impact the decision to refer a patient to rheumatology. In a multivariate model incorporating all the significant predictors but excluding anti-CCP antibody status, increasing age, being a non-smoker, higher joint counts (tender and swollen), and worse patient global score all shortened time to diagnosis (Table 4). All levels of education beyond elementary school were associated with a longer time to diagnosis, as was an annual income <$20,000 Canadian per year.
Table 4

Multivariate analysis for predictors of time to diagnosis (per month), excluding anti-CCP antibody results



95 % CI

p value



4.713, 7.555


Sex (female)


−0.466, 0.335


Ethnicity (Caucasian)


−0.048, 0.883


Education level



−0.082, 2.011


 High school


0.302, 2.115




0.197, 2.032




0.459, 2.362


Income level

 No answer


−0.330, 1.018




0.175, 1.768




−0.212, 1.208




−0.262, 1.179



 Never smoke


−1.089, −0.115




−0.829, 0.144


Swollen joint count (28 joints), per joint


−0.107, −0.046


Patient global score, per 1 unit increase on a 0–10 visual analogue scale


−0.133, −0.009


Age, per year


−0.026, −0.0002


In the logistic regression models, the odds of being diagnosed with <6 months of disease duration were increased for each swollen joint present (odds ratio (OR) 1.04, 95 % CI 1.02–1.06 per joint), each 1 mm/h elevation in the ESR (OR 1.01, 95 % CI 1.00–1.02), and decreased for patients who were either RF or anti-CCP positive compared to both factors being negative (OR 0.68, 95 % CI 0.51–0.91). Sex, ethnicity, education levels, income levels, and smoking status were not significant in these models. Findings were similar in the models, which excluded anti-CCP status (OR for each swollen joint 1.04, 95 % CI 1.02–1.07; OR for each 1 mm/h increase in the ESR 1.01, 95 % CI 1.00–1.01).


Using data from a Canadian early RA cohort, we are able to demonstrate that more severe disease activity, characterized by more swollen joints, higher acute-phase reactants, and worse patient global scores, predicts a shorter time to diagnosis. However, diagnosis in anti-CCP antibody-positive patients and those with lower income levels are delayed relative to other early RA patients. These findings confirm that our healthcare system can effectively prioritize the most severe case presentations for specialty consultation, but that patients with more indolent presentations may experience delays in diagnosis. Consistent with Kumar et al. [6], we found that positive serology results are actually associated with delays in time to diagnosis in multivariate analysis. The reasons for this unanticipated finding are unclear. One potential reason is that some triage units may delay assigning a priority to the patient for an appointment until the serology results are available, which will delay the time to diagnosis. In addition, availability of anti-CCP testing is restricted in some regions of Canada, despite its utility in diagnosis and prognosis of RA. Positive anti-CCP results could potentially increase referral to rheumatology, particularly for patients with fewer swollen or painful joints, and be in the best interest of patients for optimal outcomes. Finally, our results highlight the risk of delays in diagnosis based on an individual’s socioeconomic status, despite the universal healthcare access system in Canada.

The importance of early referral to a rheumatologist to improve patient outcomes in RA cannot be overstated. For some, the chance of obtaining disease-modifying anti-rheumatic drug (DMARD) free remission is nearly doubled if treatment can be initiated within 12 weeks of symptom onset, and the rate of joint destruction decreased by one-third [1]. As such, potential barriers to early disease recognition, referral, and diagnosis must be addressed to optimize patient outcomes. Consistent with other studies, clinical findings of joint involvement are related to earlier recognition of inflammatory arthritis. Our results are reassuring in that it does not appear that sex, age, education, ethnicity, or preexisting conditions such as obesity, mental health conditions, or painful musculoskeletal syndromes affect recognition of symptoms or referral in Canadian settings. We confirm the findings of Jamal et al. [4] that a family history of RA does not appear to increase awareness of the potential diagnosis for patients.

A delay in diagnosis may occur for multiple reasons; patients may not present to primary care providers, and/or patients or their providers may not always recognize features of RA. The rheumatologists may not be able to accommodate new RA referrals in a timely manner or may not be certain of a diagnosis within the first few visits [16]. Qualitative studies have investigated factors contributing to delays in seeking care in patients with early RA and are summarized by Stack et al. [17]. The severity, intensity, and duration of symptoms and functional disability increase the likelihood of presentation for care. Many patients attributed their symptoms to a normal part of aging or provide alternative explanations for their symptoms. There is a lack of public awareness about what RA is and what its initial symptoms and prognosis are. Referring physicians may also not appreciate that suspected RA should be a high priority referral. Finally, the quality of healthcare interactions, accessibility, and patient attitudes will impact referral for care. From the primary care providers’ perspective, disease characteristics, patient preferences, specialist access, clinical and administrative leadership, and interpersonal relationships with their patients and specialist colleagues impact time to referral [18]. Two Canadian studies have indicated that once a rheumatologist does identify early RA, there is minimal delay in starting treatment [4, 19].

Previous studies have considered disease severity, patient demographic and socioeconomic class, specialist practice characteristics, and investigation results in evaluating time to diagnosis. In one retrospective Canadian study, patients with higher baseline swollen joint counts receive treatment earlier, but age, sex, ethnicity, education, comorbidity, rheumatologist practice type, and years since physician graduation did not affect time to treatment [4]. In contrast, in a population-based study from Quebec, only 27 % of patients coded as having RA by a primary care physician were referred to rheumatology within the next 2.5–3.5 years, and the strongest predictors of shorter time to consultation were female sex, younger age, higher socioeconomic class, and the presence of greater comorbidity [7]. In Venezuela, the type of healthcare system was found to affect time to diagnosis, with delays experienced by patients initially presenting to the public system, or seen first in primary care or an orthopedist [20]. Finally, a study from a hospital-based cohort of RA patients in Spain found a median lag of 17 months from onset to first visit for RA, with a shorter time to assessment for patients with a higher number of swollen and tender joint counts, older age, home support, labor force status, marital status, and years of education [5].

The impact of fewer swollen joints and normal laboratory parameters at presentation on delay to diagnosis merits further consideration. Triage processes have favored disease activity and laboratory investigation abnormalities, likely resulting in missed indolent and seronegative cases. Generally, these characteristics are associated with better disease outcomes, but this finding highlights the importance of recognizing limitations of these processes as we try to prioritize cases for review. Indeed, it has been previously stated that normal acute-phase reactants, negative serology and normal radiographs should not delay referral if there is suspicion of inflammatory arthritis [21, 22].

Our study has some limitations. Our cohort does not capture patients outside of a specialty care model. Symptom onset is self-reported by patients and thus subject to recall bias; however, by including patients within 1 year of symptom onset, this risk is reduced somewhat. It is possible that the low frequency of comorbidities of interest affected our ability to detect statistical differences in time to diagnosis. However, having conditions such as osteoarthritis or depression likely results in increased contact with healthcare providers, with the potential for increased opportunity to discuss joint symptoms and referral. Our study has used data collected at the time of enrollment to the cohort in the analysis and did not specifically study the patient experience in accessing primary care nor the primary care decisions made in the initial workup and referral of patients to specialty care. Certainly, our findings are applicable to a universal healthcare system such as that found in Canada, and our results may not be generalizable to other models of care across the world. Future research will need to specifically address these aspects to understand delays in diagnosis and treatment initiation and how to improve public recognition of RA symptoms.

The results of this study are likely a best-case scenario, where median time from persistent joint symptom onset to seeing a rheumatologist was under 6 months, facilitated by access clinics prioritizing early RA detection. This finding strongly supports the critical importance of early access clinics to aid in rapid diagnosis and management. It also supports the need for increased education of the public and physicians as to the urgency of early RA referrals. Thus, the generalizability of the data to routine rheumatology clinics where strategies to increase the timely detection of disease [16] are not available may not be possible. Finally, the CATCH study sites include a diverse representation of academic and community practices across Canada, and referral patterns likely vary between sites. Overall though, our data reflect the general status of early RA care in Canada.

In summary, access to early RA care is dependent on critical steps of symptom recognition by the patient and their primary care provider, initiation of referral to a rheumatologist, and patient review and diagnosis by the specialist. All these steps provide intervention opportunities to assure timely, equitable, and appropriate management of RA. We did not find any evidence of discrimination in time to diagnosis based on ethnicity or underlying health comorbidities that we studied, but patients with lower income levels are at risk of delays in diagnosis. We confirm findings from previous studies, demonstrating that a higher number of swollen joints, elevated acute-phase reactants, and worse patient global scores decrease time to diagnosis. Strategies to identify patients with a less severe disease presentation and normal acute-phase reactants, such as enhanced screening of autoantibody status in primary care, need to be implemented in order to allow all patients with RA an opportunity for early management of their disease with improved outcomes.


Daming Lin provided assistance with analysis.

Conflict of interest

The CATCH study was designed and implemented by the investigators and financially supported initially by Amgen Canada Inc. and Pfizer Canada Inc. via an unrestricted research grant since inception. As of 2010, further support was provided by Hoffmann-La Roche Ltd., United Chemicals of Belgium (UCB) Canada Inc., Bristol-Myers Squibb Canada Co., Abbvie Laboratories Ltd., and Janssen Biotech Inc. (a wholly owned subsidiary of Johnson & Johnson Inc.).

Copyright information

© Springer-Verlag Berlin Heidelberg 2013