Physician treatment preferences in rheumatoid arthritis of differing disease severity and activity: The impact of cost on first-line therapy




To conduct a pilot study to identify rheumatologists' treatment preferences for first-line rheumatoid arthritis (RA) therapy and determine whether pharmacoeconomic variables modify physician choice(s).


A questionnaire describing 3 different RA scenarios was mailed to American College of Rheumatology members within 4 geographic regions of the US. Physicians were asked to identify their choice(s) of first-line therapy for each of the cases, first taking cost into consideration, second without considering the influence of cost, and third identifying the therapy that would be chosen for either themselves or a family member.


Three hundred seventy-five questionnaires out of a total of 994 (37.7%) were returned between 3/12/00 and 4/25/00. Hydroxychloroquine was the most commonly cited medication for a mild disease activity/severity presentation, and methotrexate for a moderate-to-severe disease activity/severity presentation. For the severe disease activity/severity presentation, when cost was not considered, 217 (65%) rheumatologists included new disease-modifying antirheumatic drugs (leflunomide, etanercept, and infliximab) in their choice of first-line agents; this number decreased to 47 (14%) when cost was a consideration.


Pharmacoeconomics appear to play a dominant role in rheumatologists' choice of treatment regimens, at times contrary to the physician's perception of the effectiveness of a drug. Future studies should address physician preferences in more depth with respect to cost and its various components.


Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by a symmetric erosive synovitis that commonly results in significant disability. Patients with RA fall into a clinical spectrum ranging from a slowly progressive to a rapidly progressive, aggressive course. In two-thirds of RA patients, joint destruction begins within the first year (1) and the longer the disease remains untreated, the worse the outcome (2).

Although the exact pathogenesis of RA is unknown, modestly effective therapies have been developed that at least ameliorate symptoms and delay disease progression. The goals of long-term RA treatment are: 1) to prevent synovial inflammation because its presence and persistence set the stage for significant articular damage; 2) to modify the progressive nature of disease so as to prevent erosions and joint-space narrowing, and their attendant deformities and dysfunction; and 3) to achieve the ultimate goal of “no evidence of disease” (3). In the past, a step-wise or pyramidal approach to RA therapy was standard. More recently, the paradigm has shifted. Evidence suggests that maximal success depends largely on early and aggressive medical therapy.

The selection of the most appropriate disease-modifying antirheumatic drug (DMARD) regimen is a complex clinical decision. Rheumatologists have several drugs to choose from, either alone or in combination, each with different costs, monitoring protocols, potential risks, and benefits. Several DMARDs, including hydroxychloroquine (HCQ), sulfasalazine (SSZ), gold, azathioprine (AZA), D-penicillamine (D-Pen), cyclosporine (CSA), and methotrexate (MTX) have long been used in the treatment of RA. Recently, leflunomide (LEF) has been added to this list.

The most impressive advance in the management of RA was the development of highly effective and safe tumor necrosis factor α (TNFα) antagonists (etanercept and infliximab), biologic agents that can neutralize a central proinflammatory cytokine, TNFα, and markedly improve the clinical outcomes to levels not previously seen. Despite these clinically important findings, their cost is significant and reimbursement varies greatly, depending on the age of the patient, section of the country, and third-party payer. In response to these costs, most third-party payers, including Medicaid, restrict the payment for these new biologic agents to those patients who do not demonstrate an adequate clinical response to full-dose MTX. This requirement compels physicians and their staff to spend more time completing extra paper work and phone calls. Therefore, more than ever, cost constraints and administrative burdens are likely to limit patient access to these new treatments and alter the ease with which physicians can prescribe them, especially as first-line therapy.

The primary objective of this pilot study was to identify the physician treatment patterns for first-line RA therapy and to determine whether pharmacoeconomic variables modify their choice. We also specifically examined: 1) physician preferences for use of breakthrough TNFα antagonists as first-line therapy for RA and the impact of cost on this selection; and 2) physician and geographic variables that might be associated with physician choice for differing first-line RA therapy. For this investigation, we used a study design that would not limit physicians' responses, thus generating hypotheses without introducing our own bias.


A questionnaire describing 3 different patients with RA was developed. The cases described a patient of the same age and sex, but with an initial presentation of increasing disease activity/severity of a mild, moderate, and severe nature (Figure 1). The questionnaire was mailed to rheumatologists (currently in adult-practice) selected from 4 US geographic regions: the Northeast (New York [NY]); South (Texas [TX]); Midwest (Missouri/Nebraska/Kansas [MO/NE/KS]); and West (California [CA]). Physicians were identified through the 1999 membership directory of the American College of Rheumatology. For each case scenario, rheumatologists were asked 3 questions regarding their choice of first-line therapy (defined as physicians' initial therapy choice). These questions were open-ended, which allowed the physicians to respond using monotherapy or combination therapy that could include any class of drug: nonsteroidal antiinflammatory drugs (NSAIDs), cyclooxygenase-2 specific inhibitors (COX-2), corticosteroids, DMARDs, or biologic agents. The first question specifically stated that cost was not a consideration; the second presumed that cost was a consideration. Rheumatologists were also asked their age (as a proxy for year of training) and their choice of therapy if they or a family member were the patient (to determine their optimal treatment choice). A $10 check and a self-addressed stamped envelope were enclosed with the questionnaire as an incentive to increase the response rate.

Figure 1.

Questionnaire describing 3 different rheumatoid arthritis scenarios with increasing disease activity/severity of a mild, moderate, and severe nature.

For the purpose of analysis, medications were categorized into 5 groups: NSAIDs (including aspirin compounds), COX-2, corticosteroids (PRED), traditional DMARDs (HCQ, SSZ, AZA, CSA, D-Pen, gold, MTX), and new DMARDs (LEF and TNFα antagonists [etanercept and infliximab]).

For the 3 case scenarios, responses were analyzed by comparing the number of individual medications (Chi-square test) and treatment regimens for each of the 3 questions: question (Q)1 cost is not a consideration, Q2 cost is a consideration, and Q3 the patient with this disease presentation is either the physician or a family member. Further analysis based on the geographic location and age of the rheumatologist was performed. Parts of the questionnaire were eliminated from analysis if responses were missing or unclear.


Nine hundred ninety-four questionnaires were mailed in March 2000, and 375 (37.7%) were returned between 3/12/00 and 4/25/00. The response rates by region were similar (MO/NE/KS 44.2%, NY 36.4%, CA 35.3%, and TX 31.5%). Following exclusion of incomplete or inaccurate responses, the number of qualified questionnaires was reduced to 31-34% (based on the 3 different RA scenarios). Overall analysis was performed on these responses.

Presentation with increasing disease activity/severity resulted in a steadily increasing frequency of selection of PRED, MTX, LEF and TNFα antagonists, while the frequency of selection of NSAIDs/COX-2, and HCQ decreased. The frequency of SSZ selection decreased with increasing disease activity/severity when cost was not a consideration but increased when cost was considered (Figure 2). Fewer than 10 respondents for each case scenario cited CSA, gold, or minocycline regardless of cost; and no respondent selected D-Pen as a first-line treatment.

Figure 2.

The most commonly cited medications as first-line therapy in rheumatoid arthritis scenarios with mild, moderate, and severe disease activity/severity. NSAID/COX-2 = Non-steroidal antiinflammatory drugs/cyclooxygenase-2 specific inhibitors; HCQ= Hydroxychloroquine; SSA= Sulfasalazine; MTX = Methotrexate; Anti-TNF = Anti-tumor necrosis factor (etanercept or infliximab). ▪ Cost is not a consideration ($not) □ Cost is a consideration ($) * $not versus $; P ≤ 0.02 by Chi-square test.

Medication cost did not influence the choice of most commonly (top 5) cited treatment regimens as first-line therapy for the cases of mild to moderate disease activity/severity (Tables 1 and 2). HCQ was the most commonly cited DMARD for mild disease activity/severity, and MTX for moderate disease activity/severity.

Table 1. The 5 most commonly cited treatment regimens as first-line therapy in rheumatoid arthritis scenarios with mild disease activity/severity
Treatment regimen$ not n (%)$ n (%)
  1. $ not = cost is not a consideration; $ = cost is a consideration; NSAID/COX-2 = nonsteroidal anti-inflammatory drug/cycloxygenase-2-specific inhibitors; HCQ = hydroxychloroquine; MTX = methotrexate.

#1 NSAID/COX-2+HCQ99 (29)96 (28)
#2 NSAID/COX-2+MTX52 (15)54 (16)
#3 NSAID/COX-228 (8)30 (9)
#4 NSAID/COX-2+HCQ+MTX23 (7)21 (6)
#5 MTX20 (6)24 (7)
Table 2. The 5 most commonly cited treatment regimens as first-line therapy in rheumatoid arthritis scenarios with moderate disease activity/severity
Treatment regimen$ not n (%)$ n (%)
  1. PRED = corticosteroids. For additional abbreviations, see Table 1.

#1 MTX+NSAID/COX-2+PRED45 (15)64 (21)
#2 MTX+NSAID/COX-234 (11)49 (16)
#3 MTX+PRED23 (7)30 (10)
#4 MTX17 (5)27 (9)
#5 MTX+NSAID/COX-2+HCQ18 (6)20 (6)

The most commonly cited treatment regimens for severe disease activity/severity are shown in Table 3. TNFα antagonists were included in 66% (90 of 136) of the 5 most commonly cited regimens when cost was not a factor, and no respondents cited TNFα antagonists as one of the 5 most common regimens when cost was considered (P < 0.001). When cost was not considered, 217 (65%) rheumatologists included new DMARDs (LEF = 26, etanercept = 135, infliximab = 16, etanercept or infliximab = 40) in their choice of first-line agents. This number decreased to 47 (14%) (LEF = 15, etanercept = 18, infliximab = 6, etanercept or infliximab = 8) when cost was a consideration (P < 0.001).

Table 3. The 5 most commonly cited treatment regimens as first-line therapy in rheumatoid arthritis scenarios with severe disease activity/severity
Treatment regimen/$ notn (%)Treatment regimen/$n (%)
  1. TNF = tumor necrosis factor α (etanercept or infliximab). For additional abbreviations, see Table 1.

#1 MTX+TNF+PRED36 (11)MTX+PRD+NSAID/COX-260 (20)
#2 MTX+PRD+NSAID/COX-229 (9)MTX-PRD40 (12)
#4 MTX+TNF26 (8)MTX+NSAID/COX-222 (7)
#5 MTX-PRD17 (5)MTX22 (7)

As severity of disease activity increased, rates of combination DMARD therapy as choice of first-line treatment for RA compared with rates of monotherapy increased, regardless of cost. Similarly, the proportion of physicians choosing TNFα antagonists as part of these combinations increased. However, the proportional rate of TNFα antagonist use did not increase as much when cost was taken into consideration. For the most severe case scenario, the number of physicians who preferred the combination of 3 non-TNFα antagonist DMARDs increased when cost was a consideration (Table 4).

Table 4. Percentages of physicians citing combinations of DMARDs and TNFα antagonists as part of the combination in the treatment of rheumatoid arthritis based on disease activity/severity*
Regimen useCost is not a considerationCost is a consideration
  • *

    On a background of non-steroidal antiinflammatory drugs, or cyclooxygenase-2-specific inhibitors, and/or corticosteroids. DMARDs = disease modifying antirheumatic drugs; TNF = tumor necrosis factor α.

1 DMARD735541756047
2 DMARD143545122629
3 DMARD481231123
TNF as sole DMARD62645024
TNF in 2 DMARD20598101628
TNF in 3 DMARD053410916
Total anti-TNF633570410

The use of NSAIDs or COX-2 alone was preferred by 10% of rheumatologists for mild disease activity/severity. Regardless of disease activity/severity on presentation, COX-2 selection decreased by two-thirds when cost was considered, with a reciprocal increase in the use of traditional NSAIDs.

When treating patients presenting with mild and moderate disease activity/severity, 53–56% and 3–5% of rheumatologists respectively chose various combinations of NSAIDs, COX-2, PRED, HCQ, and SSZ as initial treatment. Monotherapy was cited by nearly 75% of all rheumatologists for the treatment of mild disease activity/severity, by 60% for moderate disease activity/severity, and by 50% for the treatment of severe disease activity/severity regardless of cost.

No difference in choice of individual medications or combination regimens was noted based on geographic location or age of the rheumatologist. Physicians cited the same regimens for their own personal therapy as those employed when cost was not a consideration.


Over the last two decades there have been substantial changes in the treatment strategy of RA. This paradigm shift and its benefits have been well documented in several studies. Ward et al reported that antirheumatic medication use by patients with RA changed significantly between 1981 and 1996. Use of intramuscular (IM) gold (14%) and D-Pen (14%), which were the most commonly used DMARDs in 1981, declined significantly over 15 years (2% and 0%, respectively) (4). A Canadian study published in 1985 indicated that no patient received initial treatment with MTX until IM gold, SSZ, or HCQ failed to control the disease (5). In 1995, a national survey of 551 rheumatologists reported that monotherapy with MTX was the first DMARD chosen for aggressive RA by 83% of those surveyed, while the initial choice for slowly progressive disease was still HCQ (6). In 1998, US rheumatologists (n = 352) indicated that MTX and combination therapy were the only DMARDs with acceptable long-term effectiveness (7).

However, the development of new biologic agents, the greater appreciation of the need for early aggressive approach in disease management, and increased use of combination therapy has brought a new perspective to the treatment of RA. It has been demonstrated that RA patients who received etanercept, but not MTX, early in the course of their disease (< 3 years of disease) had significantly less evidence of radiographic progression at 2 years compared with patients receiving MTX (up to 20 mg/week) (8–10). Similarly, infliximab in combination with MTX reduces progression of structural damage significantly more than MTX alone (11–12). In addition to new DMARDs, a new generation of NSAIDs (COX-2 inhibitors [celecoxib and rofecoxib]) has recently been added to RA treatment regimens. Due to a significantly lower gastrointestinal side effect profile compared with traditional NSAIDs (13, 14), COX-2 inhibitors are increasing in popularity among physicians.

We have entered a new and exciting era in the field of rheumatology in which novel therapies offer patients hope for improved short- and long-term outcomes. Despite these clinically important findings, their cost is significant, averaging $14,000 per year for etanercept and $8,000–11,000 per year for infliximab (15). In addition, the cost of COX-2 inhibitors ($1,000–2,000/year) is significantly greater than generic NSAIDs.

Although the optimal treatment regimen for early RA has yet to be established, recent evidence strongly suggests that an early, aggressive approach results in the best possible outcome (16). Our survey demonstrates that for the first-line treatment of early RA with mild disease activity/severity a surprisingly high percentage of rheumatologists choose NSAIDs (not DMARDs) as monotherapy; and half of the rheumatologists do not prefer MTX, gold, AZA, CSA or any of the newer DMARDs (LEF, etanercept, infliximab) in this broad RA patient population. Thus, aggressive, combination and more costly DMARD therapy has not been universally accepted for use in early RA with mild disease activity/severity.

For moderate to severe disease activity/severity, our study demonstrates that despite many available DMARDs, MTX is still the most commonly selected DMARD as first-line therapy. However, when cost is considered, our findings suggest that physicians may limit their options for first-line treatment of RA, especially when considering TNFα antagonists. For severe disease activity/severity, when cost is not a consideration, TNF-based regimens are preferred by 55% of rheumatologists (decreasing to 10% when cost is a consideration), despite the relatively short market life of these medications.

The great variability in prescribing patterns for RA treatment may be explained by patient and physician preference, physician experience with specific drug classes, and physician specialty (17). Our study shows that physicians may choose a treatment regimen based largely on cost, at times contrary to their perception of a drug's effectiveness in altering the outcome of this aggressive and disabling inflammatory disorder. The cost of the medication and probably its attendant administrative burdens present themselves as other factors complicating physician selection of first-line therapy for RA.

When calculating the cost of drug therapy, it is important to consider not only the price of the drug, but also the cost of monitoring and treating the toxicities each might incur (18). The increased use of new biologic agents with or without combination DMARDs will increase the immediate drug costs but may potentially result in significant long-term savings with respect to surgical and rehabilitative costs, as well as the indirect costs of disability.

The common limitations, found in survey studies, have to be considered when examining our findings. RA presentations are highly variable, and rheumatologists may interpret these RA scenarios differently. Also, our survey does not reflect all rheumatologists because the response rate was only 38% (albeit above the average of mail survey response rates) (19). Finally, since infliximab was not available until several months after etanercept, our survey may not reflect the present practice regarding the class of TNFα antagonists.

“Cost consideration” represents the actual cost of the medication. However, indirectly it also represents the cost to the third-party payer, the cost to the patient (reimbursement policies of third-party payers such as managed care companies and government programs), or any combination of these. Future studies should address physician preferences in more depth with respect to the various components of cost. Furthermore, as the potential benefits of expensive and more effective RA regimens are better realized, the assessment of the apparent large impact of cost is warranted (i.e., individual and societal cost-benefit analyses).

This pilot study demonstrates that pharmacoeconomics appear to play a dominant role in rheumatologists' choice of treatment regimens. These data compel all of us to consider the impact of these costs on medication selection and the ultimate benefit to the patient. In addition, these data also place a burden on pharmaceutical manufacturers and third-party payers to make efforts to decrease the expense of these highly effective medications, thus improving patient access and outcome.