By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Treatment guidelines for rheumatoid arthritis (RA) now recommend early, aggressive, and persistent use of disease-modifying antirheumatic drugs (DMARDs) to prevent joint damage in all people with active inflammation, and evaluation by a rheumatologist, when possible. This research assesses whether care for RA, at a population level, is consistent with current treatment guidelines.
Using administrative billing data from the Ministry of Health in 1996–2000, all prevalent RA cases in British Columbia, Canada were identified. Data were obtained on all medications and all provincially-funded health care services.
We identified 27,710 RA cases, yielding a prevalence rate of 0.76%, consistent with epidemiologic studies. DMARD use was inappropriately low. Only 43% of the entire RA cohort received a DMARD at least once over 5 years, and 35% over 2 years. When used, DMARDs were started in a timely fashion, but were not used consistently. Care by a rheumatologist increased DMARD use 31-fold. Yet, only 48% and 34% saw a rheumatologist over 5 and 2 years, respectively. DMARD use was significantly more frequent, persistent, and more often used as combination therapy with continuous rheumatologist care. DMARDs were used by 84% and 73%, 40%, and 10% of people followed by rheumatologists continuously and intermittently, internists, and family physicians, respectively (P < 0.001). NSAID use, physiotherapy, and orthopedic surgeries were similar across these 4 care groups.
RA care in the British Columbia population was not consistent with current treatment guidelines. Efforts to educate family physicians and consumers about the shift in RA treatment paradigms and to improve access to rheumatologists are needed.
Rheumatoid arthritis (RA), a chronic inflammatory arthritis, leads to high levels of pain, functional disability, decreased ability to work, and premature mortality (1). Standards of care have been established for its treatment (2–5). Disease-modifying antirheumatic drugs (DMARDs) reduce joint destruction and long-term physical disability (6–8). Delays, as brief as 3 months, in instituting DMARDs have been associated with poorer long-term outcomes, including greater physical disability (9–11) and increased radiologic damage (12–14). Therefore, treatment with DMARDs is recommended in all RA patients with active disease (3), and treatment paradigms have shifted towards early, aggressive, and persistent use of DMARDs with the aim of eradicating inflammation, thereby preventing joint damage (15–17).
Guidelines recommend early referral to a rheumatologist when a diagnosis of RA is suspected, for definitive diagnosis and early treatment with DMARDs (4, 5), and continued involvement of a rheumatologist to assess response to treatment and review the treatment plan (3). When rheumatologists are involved, quality of RA care (18) and outcomes (19–21) are improved, at no increased cost (22). Rarely is compliance with recommended guidelines assessed at the population level; however, this is essential to determine the quality of health service provision. In this study, we examined care received by the RA population of the province of British Columbia, Canada as captured by provincial health administrative databases, and compared it with recommended guidelines (2–5), focusing on DMARD treatment and specialist care.
The specific objectives were to evaluate whether DMARD use at the population level meets recommended guidelines by measuring the proportion of patients treated with DMARDs, the timeliness of starting DMARDs, the consistency of long-term DMARD use, and the frequency of use of combination therapy; to measure the proportion of the RA population treated by a specialist (rheumatologist or internist) and evaluate the impact on process of care; and to identify the determinants of DMARD use and care provided by a specialist.
PATIENTS AND METHODS
Administrative billing data, collected by the Ministry of Health of British Columbia for the reimbursement of physician visits, was used to identify all individuals who received care for RA between January 1996 and December 2000. Applying the case definition of MacLean et al (18), individuals were selected as RA cases if they had at least 2 physician visits more than 2 months apart with an RA diagnostic code (International Classification of Diseases, Ninth Revision code 714.X), thus excluding initial impressions of RA not confirmed on second visits and transient inflammatory arthritis lasting <2 months. Cases of juvenile-onset rheumatoid arthritis were included. To improve specificity, we excluded individuals with at least 2 visits, subsequent to the second RA visit, with diagnoses of other inflammatory arthritides (systemic lupus erythematosus, other connective tissue diseases, psoriatic arthritis, ankylosing spondylitis, and other spondylarthropathies) and those where a diagnosis of RA by a nonrheumatologist was not confirmed when the individual saw a rheumatologist. RA cases were identified as incident cases for 1996–2000 if they had no prior visit for RA since 1990.
Data on all provincially-funded health services used from January 1990 to December 2000 were obtained from administrative databases of the Ministry of Health of British Columbia. This included data on all physician visits, up to 10 physiotherapist visits per year, and all hospitalizations. Data were obtained on all prescription medications dispensed by pharmacists, from January 1996 to December 2000, for all RA cases, regardless of age or funding source.
No personal identifying information was provided. Procedures were compliant with British Columbia's Freedom of Information and Protection of Privacy Act. The study received ethics approval from the University of British Columbia.
DMARD use was a binary variable defined as any prescription of (hydroxy)chloroquine, sulfasalazine, gold, methotrexate, cyclosporine, azathioprine, cyclophosphamide, chlorambucil, or leflunomide. D-penicillamine was used very rarely in 1996–2000, and biologic agents had not yet received regulatory approval in British Columbia. Consistency of DMARD use was assessed by measuring the percentage of observation time when DMARDs were prescribed, starting from their first use. The duration of DMARD prescriptions was calculated according to the date dispensed and the dose and amount prescribed. Because physicians might decrease the dose after the DMARD was dispensed and the resultant decrease might appear as discontinuation, gaps of 4 months or less between the end of one prescription and the beginning of the next prescription dispensed for the same medication were considered as part of the same treatment course. Four months was selected because 99% of DMARD prescriptions lasted ≤4 months and therefore, this allowed up to a 50% dose reduction. The dose of DMARD used was not analyzed as a marker of aggressive use in the assessment of guidelines, because of the uncertainty around the actual daily dose consumed, when using data from administrative databases to assess the dose of medications that are frequently adjusted after prescription.
Combination therapy was defined as having received ≥2 DMARDs with overlapping prescription durations. To ensure that discontinuation of one DMARD and starting of a second DMARD as single therapy was not erroneously considered combination therapy, DMARDs not prescribed simultaneously were considered used in combination only if the first DMARD prescribed was dispensed again after starting the second one.
Physician care for RA was classified into care provided by rheumatologists, internists, or family physicians. Family physician care meant the individual had never been seen by a rheumatologist or by an internist for RA. Because in British Columbia internists practice as consultants, requiring a referral from a family physician, care provided by rheumatologists and internists was combined for analyses requiring a binary variable and was referred to as care provided by a specialist.
Care provided by a rheumatologist was further subdivided into continuous versus intermittent, using the same methodology as used by Ward et al (19), where continuous care was defined as at least 1 rheumatologist visit per 6-month observation period. All rheumatologists practicing in British Columbia were identified using the British Columbia Society of Rheumatology criteria. Rheumatologist consultations provided through traveling services to 22 outlying areas were also included.
Assessment of other quality of care indicators.
Corticosteroid use was assessed as intraarticular and oral administration. All oral preparations were converted to prednisone equivalent dose. Dosages >40 mg/day and intravenous pulses were excluded to prevent steroid pulses and use for extraarticular features from skewing calculations. When calculating steroid use duration, lapses ≤4 months between consecutive prescriptions were considered part of the same steroid course, to account for reductions in dosage ≤50%, as described for DMARDs. Use of nonsteroidal antiinflammatory drugs (NSAIDs) was assessed.
Physiotherapy, massage therapy and visits to chiropractors, osteopaths, and naturopaths were assessed, as they are partially provincially funded, but occupational therapy was not assessed. Physiotherapy did not include hospital-affiliated programs, likely underestimating overall rates, but nonetheless allowing comparison across groups. Surgical management of RA was assessed using hospitalization data and procedure codes for reimbursement of surgeon services. Surgical procedures assessed included arthroplasties (hip, knee, ankle, shoulder, elbow, and wrist), arthrodesis (ankle, wrists, cervical-spine, and other joints), reconstructive surgeries (acromioplasty and rotator cuff reconstruction, distal ulnar resection and radioulnar instability reconstruction, and reconstructive surgeries of hands and feet), synovectomies (shoulder, elbow, wrist, hand, and tendons), and tendon repairs (hands and feet).
To evaluate care at the population level, the proportion of RA cases treated at least once with any DMARDs and with combination therapy, as well as the proportion of RA cases seen at least once by a rheumatologist or an internist versus those followed by family physicians, were calculated. Proportions were calculated over 5-, 2-, and 1-year periods using all prevalent cases. Data from the most recent years available (1996–2000, 1999–2000, and 2000) were chosen to be representative of the most up to date care. Consistency of DMARD use was also calculated, as described above.
The determinants of DMARD use and of specialist care were identified using multivariate logistic regression analysis using a forward stepwise method with backward elimination. For DMARD use, the dependent variable was whether DMARDs were used at least once in 1996–2000. Specialist care (combining rheumatologist and internist care) was assessed in a similar manner. Explanatory variables included age, sex, geographic area, and, as a proxy of socioeconomic status, whether people received subsidization for payment of their provincial health insurance premiums. Additionally, type of physician providing RA care was included in the regression analysis for DMARD use. Variables were selected in the final model if P < 0.05.
To evaluate timeliness of care, incident RA cases from 1996–2000 were used to measure time from first RA diagnosis to starting DMARDs and to first rheumatologist visit. Cumulative use of DMARDs stratified by type of physician care was also measured, using Kaplan-Meier survival analysis.
To evaluate the impact of specialist care on process of RA care between 1996 and 2000, the RA population was divided into 4 groups according to whether people had seen a rheumatologist continuously, a rheumatologist intermittently, an internist, or a family physician. The 4 groups were compared for various care indicators, including the use of DMARDs, use of combination therapy, the consistency of DMARD use, use of NSAIDs and corticosteroids (including dose and duration), use of physiotherapy and complementary alternative therapies, visits to orthopedic surgeons, and orthopedic surgeries for RA. To evaluate whether comorbid diseases influenced the likelihood of specialist care, the 4 groups were compared for the number of unique diagnostic codes on physician visits, using a modification of the Charlson index developed for analysis of administrative data from outpatient physician visits (23–25). Comparisons across the 4 groups were performed using chi-square analysis for categorical variables, comparison of means using Student's t-test for continuous variables with normal distribution, and comparison of medians using Wilcoxon's rank sum test.
Between January 1996 and December 2000, 27,710 people met the criteria for RA. This yields a prevalence of 0.76% among the noninstitutionalized population of British Columbia over age 16 for 1999. The mean ± SD age was 64.1 ±17.0 years, 67% were women, and 37% had received some subsidization for their health insurance premiums.
DMARD use at the population level.
Only 43% of the entire RA population had used a DMARD at least once over 5 years, 35% over 2 years, and 31% over 1 year. Use of DMARDs was greatly influenced by specialist care; 76% of people seen at least once by a rheumatologist used a DMARD over a 5-year period, compared with 40% for internists and 10% for family physicians. Results did not differ significantly when looking at 2- and 1-year periods (Table 1).
Table 1. Proportion of the rheumatoid arthritis population taking disease-modifying antirheumatic drugs over specific time periods, according to specialty of physician providing care*
5 years (1996–2000)
2 years (1999–2000)
1 year (2000)
Values are the percentage (number of patients/total number of patients).
Of the people seen by a rheumatologist, those who switched to receiving RA care from their family physician were less likely to continue on a DMARD. Of the people who saw a rheumatologist in 1996–1998 but did not see one in 1999–2000 (i.e., switched to family physician care) (n = 3,333), only 34% received a DMARD, compared with 76% of those who continued to see a rheumatologist (relative risk 2.22, 95% confidence interval 2.11–2.32). This indicates that lack of followup with a rheumatologist is associated with suboptimal DMARD use. The number of people (n = 69) who switched from rheumatologist to internist care was too small to analyze.
Timeliness of starting DMARDs.
When DMARDs were prescribed, they were started in a timely fashion. In 30% of cases, DMARDs were started prior to or at the time of the first visit with a diagnosis of RA. The remainder of cases were started within 2.2 months (median) of first RA diagnosis (interquartile range [IQR] 0.76–6.6).
Consistency of DMARD use and use of combination DMARDs
Once prescribed, DMARDs were used for 69% of the remainder of the observation period. DMARDs were prescribed in combination at least once in 11% over 5 years, 9% over 2 years, and 7% over 1 year.
Determinants of DMARD use.
In the multivariate regression analysis assessing probability of DMARD use after controlling for regional variation, the strongest determinant was specialist care (Table 2). RA cases followed by rheumatologists were 31 times more likely, and by internists 5 times more likely, to receive DMARDs than those who received all their RA care by family physicians. Other significant determinants were sex and socioeconomic status.
Table 2. Multivariate regression analysis evaluating the probability of DMARD use and specialist care after controlling for regional variation*
DMARD = disease-modifying antirheumatic drug; OR = odds ratio; 95% CI = 95% confidence interval; Specialist care = rheumatologist and internist combined.
Over 5, 2, and 1 years, only 48%, 34%, and 26%, respectively, visited a rheumatologist at least once (Table 3), and 46%, 59%, and 64%, respectively, received all their RA care by their family physician. The remainder saw internists. Care by rheumatologists varied across regions, ranging from 27% to 54%. When care by rheumatologists and internists was combined, there was less variation, with specialist care ranging from 39% to 59%. Of the people seen by a rheumatologist in 1996–2000, 70% were followed intermittently. These people missed a rheumatologist visit over a 6-month period, on average, 53% of the time.
Table 3. Proportion of rheumatoid arthritis (RA) population receiving RA care from a specialist versus a family physician over 5-year, 2-year, and 1-year periods.
5 years (1996–2000)
2 years (1999–2000)
1 year (2000)
* Values are the % (number). NA = not applicable.
Did not see a physician
Determinants of specialist care.
Results of the multivariate regression analysis assessing the probability of specialist care, after controlling for regional variation, revealed that women, people of younger age, and people of higher socioeconomic status were more likely to receive specialist care (Table 2).
Timeliness of specialist care after RA diagnosis.
When family physicians referred RA cases, referral was timely. Thirty-five percent received a diagnosis of RA prior to their first rheumatologist visit. In these people, a rheumatologist was seen within 2.2 months (median) of first RA diagnosis (IQR 1.0–5.5).
Impact of specialist care on treatment with DMARDs.
People seeing a rheumatologist continuously were more likely to receive a DMARD than those seeing a rheumatologist intermittently (84% versus 73%, respectively; P < 0.0001), versus 40% for internists and 10% for family physicians (Table 4). The greatest impact of continuous versus intermittent rheumatologist care was on consistency of DMARD use. Once started, DMARDs were used for 83% of the remainder of the observation period for continuous versus 66% for intermittent rheumatologist care (P < 0.0001). The latter group was nearly identical to internist and family physician care. DMARDs were started earlier in people seeing rheumatologists (Table 4). DMARDs were used in combination in 30% for continuous rheumatologist care, compared with 18% for intermittent rheumatologist care, 8% for internist care, and 1% for family physician care (P < 0.0001). Frequency of individual type of DMARD used varied according to specialty of physician providing RA care (Figure 1).
Table 4. Impact of specialist care on process of care-treatment with DMARDs, corticosteroids, and NSAIDs during a 5-year period*
Rheumatologist continuously n = 3,982
Rheumatologist intermittently n = 9,291
Internist n = 1,656
Family physician n = 12,781
Except where indicated, values are the percentage of patients in each care group who received treatment indicated during the 5-year period (1996–2000). DMARDs = disease-modifying antirheumatic drugs; NSAIDs = nonsteroidal antiinflammatory drugs; IQR = interquartile range; 95% CI = 95% confidence interval.
Percentage of observation time spent taking DMARDs, calculated from the time of first DMARD prescription to the end of the observation period in people treated with DMARDs. Mean values are presented for each care group.
Time from the first rheumatologist visit to the first DMARD prescription, in patients with DMARDs started after first rheumatologist visit.
Percentage of observation time spent taking corticosteroids in those who received corticosteroids, calculated from the time of first diagnosis of rheumatoid arthritis until the end of the observation period.
Calculated as prednisone equivalent in people treated with corticosteroids, median value for each care group.
In those who received corticosteroids (calculated as prednisone equivalent), median value for each care group.
The cumulative use of DMARDs according to time from RA diagnosis, stratified by specialty of physician providing RA care, shows that people followed by rheumatologists continuously were significantly more likely to receive DMARDs (P < 0.001, by log rank test) (Figure 2).
Impact of specialist care on corticosteroid and NSAID use.
Intraarticular injections of corticosteroids were more commonly used by rheumatologists than other groups. People followed by rheumatologists were more likely to have received oral corticosteroids, spent more time on corticosteroids, and used a greater total cumulative dose, but used a lower average daily dose. Overall, 20% of the people who did not receive DMARDs received oral corticosteroids, ranging from 15% for people followed by a family physician to 37% for people followed by a rheumatologist continuously. Despite the marked variation in DMARD use, dispensing of NSAIDs was clinically similar across the 4 care groups, ranging from 79% to 89%.
Impact of specialist care on use of nonpharmacologic treatments.
For rheumatologists, internists, and family physicians, physiotherapy was used by 47%, 48%, and 46%, respectively. Chiropractors, massage therapists, and naturopaths were seen by 27%, 17%, and 7% of the RA population, respectively, without any significant difference across care groups.
Impact of specialist care on surgical management of RA.
The surgical management of RA was similar across care groups. For rheumatologists, internists, and family physicians, respectively, 37%, 40%, and 35% saw an orthopedic surgeon; 16%, 12%, and 13%, respectively, had an orthopedic surgical procedure of any type; and 7%, 5%, and 7%, respectively, had an arthroplasty over the 5-year period assessed.
There was little difference in the severity or the number of comorbid diseases across the 4 care groups. For continuous and intermittent rheumatologist care, internist, and family physician care, the mean ± SD Charlson's index was 2.49 ± 2.06, 2.78 ± 2.49, 2.84 ± 2.45, and 2.72 ±2.39, respectively; and the mean number of unique diagnoses was 25, 25, 26, and 24, respectively.
This study assessed patterns of care received by all patients with RA in the province of British Columbia, Canada, where access to health care is universal. Although our study did not assess all aspects of current RA treatment guidelines, our results clearly point to important gaps in care. Of particular concern, the use of DMARDs, essential for the treatment of active RA, was unacceptably low, with less than half the RA population having ever received a DMARD over 5 years. When dispensed, DMARDs were started in a timely fashion, but were not used consistently and combination therapy was infrequent. Care by a rheumatologist greatly increased DMARD use. Unfortunately, less than half of the patients saw a rheumatologist.
Use of DMARDs was not consistent with current clinical guidelines when internists or family physicians were the sole providers of RA care. Rates of DMARD use at 40% and 10%, respectively, are much lower than would be expected, even when taking into consideration an RA population with milder disease. Lack of followup with a rheumatologist (people who switched from rheumatologist to family physician care for their RA) was associated with suboptimal DMARD use. Continuous, rather than intermittent, involvement of rheumatologists improved compliance with guidelines. DMARDs were used most frequently, most consistently, and most often in combination with continuous rheumatologist care. The similarly high rates of NSAID use and physiotherapy across the 4 care groups indicate that most RA patients followed by family physicians had active disease, which, according to current guidelines, warrants DMARD use. Finally, the similar rate of orthopedic surgery indicates that some of the patients followed by family physicians had sufficiently severe disease to require surgery, including arthroplasty.
The results of our study suggest that the paradigm shift to early, aggressive, and sustained use of DMARDs, well known to rheumatologists, has not reached family physicians, as evidenced by their frequent use of NSAIDs and physiotherapy but infrequent use of DMARDs, a pattern reminiscent of the pyramid approach. Yet, family physicians followed at least half of the RA population. There is an urgent need for continuing medical education efforts, aimed at family physicians, emphasizing the necessity of DMARD treatment for all RA patients with active disease. Targeting education at consumers directly may be a more effective approach. Furthermore, the surprisingly small proportion of RA population receiving care by rheumatologists, despite universal access to health care, suggests a need for greater rheumatologist manpower and access.
The prevalence of RA in our sample (0.76%) was consistent with epidemiologic studies. In a review, prevalence rates ranged from 0.5% to 1.1%, with an average of 0.8% (26). The concordance of our prevalence rate suggests that the under treatment of RA observed in our cohort was not due to unintentional inclusion of large numbers of non-RA cases. Although some non-RA cases may have been included, they would not be sufficient in number to explain the large gap in DMARD use observed.
To our knowledge, this is the first published report assessing whether treatment guidelines for RA are applied at the population level. Our rate of DMARD use is consistent with that observed in people over age 65 in the Canadian province of Ontario (27). Our results also confirm those of nonpopulation-based studies comparing process and outcome of RA care according to specialist involvement. MacLean et al (18) also found process of care for RA to be more favorable when rheumatologists were involved, although pattern of DMARD use and other treatments was not assessed. Ward et al (19) reported that continuous rheumatologist involvement, compared with intermittent, was associated with more intense DMARD treatment. Two studies on the same patient population reported less favorable outcomes (20) and decreased likelihood of treatment with DMARDs and prednisone (21) when care was provided by nonrheumatologists rather than rheumatologists. These studies in conjunction with ours point to the need for continued rheumatologist involvement on at least a semiannual basis.
Potential limitations of our study deserve comment. They include those inherent to data from administrative databases. Uncertainty around diagnosis accuracy is the main limitation of studies identifying cases from administrative databases. Most studies in RA used case definitions based on a single visit. We used the strictest published case definition, based on 2 visits at least 2 months apart (18), which was validated against self report of a physician diagnosis of RA, yielding a positive predictive value of 0.92 (28). We further improved specificity with additional exclusions. Our data on use of DMARDs refer to medications dispensed rather than prescribed or consumed. Our analyses do not control for differences in disease severity. It is probable that family physicians treat milder RA, explaining part of the lower rate of DMARD use. However, 10% DMARD use is lower than expected, even in milder disease. Given that family physicians follow half of the RA population, one would expect more than 10% to have sufficiently active and severe disease to require DMARDs. Furthermore, the similar rate of orthopedic surgery across care groups suggests that some of the patients followed by family physicians had severe disease. Finally, while guidelines recommend that all RA patients with active inflammation be treated with DMARDs, the high rate of NSAIDs and physiotherapy use suggests that the disease of people followed by family physicians was active.
Strengths of our study include the involvement of an entire population of RA, in the context of a universal health care system, where no physicians can opt out, as well as the accurate and complete capture of health services use from administrative databases, including all medications, regardless of age and source of funding.
In conclusion, data from our population-based cohort of RA with universal health care access indicate that patterns of RA care are not consistent with current treatment guidelines and point to important gaps in care. More specifically, use of DMARDs and care by rheumatologists were inappropriately low. While none of the physician care groups had universal use of DMARDs, gaps in DMARD treatment were greatest in people whose RA care was provided solely by family physicians, which represented approximately half of the RA population. RA care was improved when rheumatologists were involved continuously rather than intermittently. Given that DMARDs are essential in the treatment of RA and have been demonstrated to impact long-term joint damage and physical function, the findings are alarming. Efforts to educate family physicians and consumers about the shift in RA treatment paradigms and to improve access to rheumatologists are needed urgently.
We would like to thank the Ministry of Health Services of British Columbia for providing data on health care utilization, as well as for their support in the funding application and in conducting this research.