The treatment of rheumatoid arthritis (RA) has. Article

Article Annals of Internal Medicine Treatment of Very Rheumatoid Arthritis With Symptomatic Therapy, Disease-Modifying Antirheumatic Drugs, or Biologic Agents A Cost-Effectiveness Analysis Axel Finckh, MD, MS; Nick Bansback, MS; Carlo A. Marra, PharmD, PhD; Aslam H. Anis, PhD; Kaleb Michaud, PhD; Stanley Lubin, MD; Marc White, PhD; Sonia Sizto, BA; and Matthew H. Liang, MD, MPH Background: Long-term control or remission of rheumatoid arthritis (RA) may be possible with very early treatment. However, no optimal first therapeutic strategy has been determined. Objective: To assess the potential cost-effectiveness of major therapeutic strategies for very early RA. Design: Decision analytic model with probabilistic sensitivity analyses. Data Sources: Published data, the National Data Bank for Rheumatic Diseases, and actual 2007 hospital costs. Target Population: U.S. adults with very early RA (symptom duration 3 months). Time Horizon: Lifetime. Perspective: Health care provider and societal. Intervention: 3 management strategies were compared: a symptomatic or pyramid strategy with initial nonsteroidal anti-inflammatory drugs, patient education, pain management, and low-dose glucocorticoids, and disease-modifying antirheumatic drugs () at 1 year for nonresponders; early DMARD therapy with methotrexate; and early therapy with biologics and methotrexate. Outcome Measures: Cost per quality-adjusted life-year (QALY) gained. Results of Base-Case Analysis: By reducing the progression of joint erosions and subsequent functional disability, both early intervention strategies increase quality-adjusted life more than the pyramid strategy and save long-term costs. When the cost of very early intervention is factored in, the cost-effectiveness ratio of the early DMARD strategy is $4849 per QALY (95% CI, $0 to $16 354 per QALY) compared with the pyramid strategy, whereas the benefits gained through the early biologic strategy come at a substantial incremental cost. The early DMARD strategy maximizes the effectiveness of early and reserves the use of biologics for patients with more treatment-resistant disease of longer duration, for which the incremental benefit of biologics is greater. Results of Sensitivity Analysis: The early biologic strategy becomes more cost-effective if drug prices are reduced, risk for death is permanently lowered through biologic therapy, patients experience drug-free remission, responders can be selected before therapy initiation, or effective alternative antirheumatic agents are available for patients for whom several biologics have failed. Limitations: Data on the long-term effect of very early therapeutic interventions on the natural progression in disability and joint erosions are limited. The study considered only tumor necrosis factor inhibitors and not the newer biologics. Conclusion: According to the most objective measures of RA progression, very early intervention with conventional is cost-effective. The cost-effectiveness of very early intervention with biologics remains uncertain. Ann Intern Med. 2009;151:612-621. For author affiliations, see end of text. www.annals.org The treatment of rheumatoid arthritis (RA) has changed dramatically in the past decade with the introduction of new biologic therapies, particularly tumor necrosis factor inhibitors. Growing evidence suggests a benefit of aggressive treatment very early in the course of the disease and the existence of a critical therapeutic window of opportunity within which antirheumatic therapy is more effective and has a more durable See also: Print Editors Notes... 613 Editorial comment....668 Web-Only Appendix Conversion of graphics into slides effect than later in the course of the disease (1, 2). Very early initiation of disease-modifying antirheumatic drug (DMARD) or biologic therapy has a demonstrated prolonged benefit on RA progression and may permanently modify the disease to a milder course (1, 3). Emerging data also suggest that if RA is detected and treated very early, long-term control or remission may be possible and continuing treatment may not be necessary (4 6). Although antirheumatic therapy should be initiated as early as possible after disease onset, the optimal strategy has not been determined. treatment with biologics seems to be superior to treatment with conventional in terms of hard outcomes, such as radiographic structural joint damage, but not necessarily in terms of disease activity (7). are generally prescribed only after therapy with 1 or more conventional has failed. However, randomized, controlled trials (RCTs) have demonstrated improved efficacy of biologics in early disease (8 612 2009 American College of Physicians

Cost Benefit Analysis for Treating Very RA Article 11), and the pharmaceutical industry promotes early initiation of biologic therapy for RA. Because early therapy may change the natural course of the disease (4), biologics could be most useful in early RA (4, 12, 13). Although evidence supports the cost-effectiveness of biologic therapy in DMARD-resistant severe RA (14, 15), its long-term cost-effectiveness in very early RA is unknown. The diagnosis of RA can be problematic very early in the course of disease (16). An early diagnosis carries the risk for prescribing expensive and potentially toxic drugs to patients who might have spontaneous remission. Patients may incur time lost from work because of treatment or drug-related toxicities or may die because of treatment (13). In addition, early antirheumatic therapy increases treatment costs; in particular, biologics cost up to 10 times more than conventional. However, patients may benefit from early therapy, endure fewer disability days, have less productivity loss, require fewer days in the hospital, and need fewer subsequent joint replacements (13). It is therefore important to consider the total longterm costs associated with various therapeutic strategies. Health policy analyses use the incremental cost effectiveness ratio (ICER) to compare a particular treatment or strategy with competing treatment strategies to inform the allocation of health care resources (17). To determine whether the increased costs of treatment strategies that involve biologics in very early RA are warranted by health gains, we assessed the cost-effectiveness of major therapeutic strategies by examining different scenarios based on emerging data. METHODS Decision Model Structure Our base-case analysis considers adults with very early RA, operationally defined as symptoms of less than 3 months in duration a definition commonly used in the literature for very early RA (18, 19). The classification criteria for RA require that signs and symptoms be present for at least 6 weeks (20), and studies of patients with early RA suggest a median duration of illness of 11 to 23 weeks at diagnosis (21, 22). However, RA cannot be reliably diagnosed at an early stage in all patients with inflammatory polyarthritis. A significant proportion of patients with undifferentiated arthritis will have spontaneous remission or receive diagnoses of other conditions (23, 24). Evolving diagnostic techniques, such as biomarkers (25) or improved clinical prediction rules for identifying which patients who present with undifferentiated polyarthritis will progress to RA, enable individualized decisions about initiation of treatment with (23, 26). We used an individual sampling model (27) to track the course of disease over time, with changes in important variables (principally radiographic joint damage and functional disability) at time points at which events (toxicities, treatment initiation, treatment discontinuation, or death) Context It remains unclear whether the optimal treatment strategy for early rheumatoid arthritis (RA) is to step care from nonsteroidal anti-inflammatory drugs and other symptomatic treatments to glucocorticoids and then to diseasemodifying antirheumatic drugs () only after 1 year, or to use more aggressive therapy early after diagnosis. Contribution The investigators used available data to estimate the cost-effectiveness of initial treatment strategies for RA. Compared with stepped care, early or early biologic agents seem to improve outcomes, but only do so at a cost per year of life saved that society typically considers acceptable. Caution Assumptions were based on limited data on early RA treatment with and biologics. The Editors occur. We estimated the expected lifetime costs and benefits for U.S. patients, expressed as quality-adjusted lifeyears (QALYs), from both a health care provider and a societal perspective. Our models explore 5 conceivable strategies, but we focus on the 3 most plausible and clinically relevant strategies for very early RA. Although these may vary in actual practice, we simplified the choices to model the problem. In the pyramid strategy, initial nonsteroidal anti-inflammatory drugs, patient education, joint protection and therapeutic exercise, pain management, and low-dose glucocorticoids are administered as needed, followed by at 1 year for nonresponders. This is still the most common pattern of therapy because of delayed consultation and diagnosis (28 30). In the early DMARD strategy, conventional (such agents as leflunomide, sulfasalazine, hydroxychloroquine, or methotrexate) are given within 12 weeks of symptom onset. Methotrexate is the current standard of care in U.S. rheumatology (30) and also has the most evidence for its effectiveness in very early RA (6). The early biologic strategy involves very early use of biologics, defined as use of 3 sequential tumor necrosis factor inhibitors, in combination with methotrexate. In the first 2 strategies, patients initially receive conventional ; those who do not respond to therapy with 3 conventional, alone or in combination, then receive agents from the third strategy. We modified a published model (31) for this analysis. We tracked hypothetical patient cohorts through the model, one at a time, in 6-month cycles from symptom onset until death. The exact route a simulated patient takes through the model depends on the therapeutic strategy used and the patient s www.annals.org 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 613

Article Cost Benefit Analysis for Treating Very RA disease characteristics (type of disease progression, age, sex, education level, race, income, duration of disease, comorbid conditions, baseline and current Health Assessment Questionnaire [HAQ] Disability Index score, and type and number of previously received ). Figure 1 shows model pathways. Figure 1. Model schematic. Treatment strategy Disease course Clinical presentation: early RA Disease states Outcomes Pain relief for 12 mo * Treat with pain relief for 12 mo* Spontaneous remission Slow progression Rapid progression Excellent response Good response Moderate response No response Death HAQ functional disability score Eroded joints Quality of life Cost Death Patients can change disease state every 6 months, depending on the underlying disease type and the treatment strategy received. The time spent in a particular disease state determines the probability of transition and the ultimate outcome. DMARD disease-modifying antirheumatic drug; HAQ Health Assessment Questionnaire; RA rheumatoid arthritis. * See Appendix (available at www.annals.org) for discussion of this strategy. We assume that patients follow 1 of 3 disease courses: drug-free remission, a mild course with slow disease progression, or rapid disease progression. At presentation, we assume that a patient s future course cannot be predicted. The patients disease characteristics are representative of the U.S. population with RA (32). The model synthesizes data from RCTs (33); the National Data Bank for Rheumatic Diseases (NDB), a longitudinal database of semi-annual, self-reported patient data (32); and other literature. Details on disease progression, treatment effect, estimates for QALYs, and costs are provided in the following sections. Table 1 (34 45) describes input variables, and the Appendix (available at www.annals.org) details our methods. We used R project statistical software (www.r-project.org) for all decision analyses. Disease Progression We modeled the course of the disease by using functional disability and radiographic evidence of structural joint damage. Functional disability, measured by the HAQ (46), is a primary outcome in most RA clinical studies. Structural joint damage can be assessed by several radiographic scoring methods. We used the percentage of the maximum score of each measure to include all sources of information (39). We applied the rate of disease progression to each disease course pattern. Because these rates are often unknown over an extended period, we first assumed that all patients present with a HAQ score of 1 and no radiographic damage. We then used the literature as a guide for the mean progression rates in each disease course. Finally, we calibrated the overall progression for all disease courses. We calibrated the HAQ with data from the NDB that show increases during early disease to an annual HAQ score progression of around 0.03 in patients with established disease, similar to previous estimates (47). We assumed that radiographic damage increased linearly over time and calibrated it with rates of approximately 2% per year (48). We computed the relationship between radiographic damage and HAQ score. Estimates indicate that a 1 percentage point change in radiographic progression is equivalent to a 0.04 change in HAQ score (38). However, the effect of radiographic progression on functional disability in early phases of disease is only experienced 5 years later (1, 3). Treatment Effect We categorized each patient s response to treatment as excellent (low residual disease activity), good, moderate, or none. A meta-analysis of DMARD therapies and evidence from early RA trials informed the rates (1, 4, 8 11). Studies (33) indicate that response rates decrease with greater disease duration and number of previous therapies. We incorporated the effect of treatment on radiographic progression in 2 ways. First, we used evidence that early DMARD treatment reduces the future rate of long- 614 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 www.annals.org

Cost Benefit Analysis for Treating Very RA Article Table 1. Key Model Parameters and Assumptions Variable Value Distribution Description (Reference) Baseline characteristics Women, % 79 Fixed Sampled from NDB (32) Age, y 50 Fixed Sampled from NDB (32) Disease duration, mo 3 Fixed Assumption Function* 1 Fixed Assumption Radiographic joint damage, %of 0 Fixed (34) maximum score Disease course, % Dirichlet Assumption (Appendix, available at www.annals.org) Spontaneous remission 10 Slow progression 70 Rapid progression 20 Short-term treatment effect, % Pyramid strategy Dirichlet Excludes patients with spontaneous remission assumption Excellent response 0 Good response 0 Moderate response 0 No response 100 DMARD strategy Multivariate normal Values shown are for a patient with baseline characteristics and decrease with longer disease duration. Low disease activity state is estimated by using multipliers based on excellent, good, or moderate response rates (4, 11, 33, 35). Excellent response 23 Good response 21 Moderate response 23 No response 33 biologic strategy Multivariate normal Values shown are for a patient with baseline characteristics and decrease with longer disease duration. Low disease activity state is estimated by using multipliers on the basis of excellent, good, or moderate response rates (4, 11, 33, 35). Excellent response 58 Good response 20 Moderate response 6 No response 16 Long-term treatment effect Reduction in the progression of joint damage, based on disease state or type of treatment, % Excellent response** 100 Dirichlet Compared with no response (36) Good response 100 Dirichlet Compared with no response (36) Moderate response 91 Dirichlet Compared with no response (36) No response 0 Dirichlet Compared with no response (36) DMARD or biologic strategy 33 Beta For subsequent 5 years only (1) Reduction in the progression of functional disability, %* Excellent response** 100 Dirichlet Based on a combination of calculations from NDB (32), improved and calibrated on the basis of literature (4). Values shown for patient with baseline characteristics. Good response 69 Dirichlet Based on a combination of calculations from NDB (32), improved and calibrated on the basis of literature (4). Values shown for patient with baseline characteristics. Moderate response 11 Dirichlet Based on a combination of calculations from NDB (32), improved and calibrated on the basis of literature (4). Values shown for patient with baseline characteristics. No response 0 Assumption Duration of treatment retention, mo Pyramid strategy 12 Assumption DMARD strategy 55 Multivariate normal Value shown for patient with baseline characteristics. Values change with disease duration, number of treatments attempted, and disease severity. Discontinuation rates are linked to type of therapeutic response better responders are more likely to continue treatment (32, 37). biologic strategy 43 Relationship between radiographic and functional progression Change in HAQ per percentage 0.04 Normal Relationship assumed to be linear (38) change in radiographic damage Lag time for worsening HAQ score from radiographic damage, y 5 Normal (39) Continued on following page www.annals.org 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 615

Article Cost Benefit Analysis for Treating Very RA Table 1 Continued Variable Value Distribution Description (Reference) Health utilities EQ-5D utility 0.76 Multivariate normal Value shown for patient with baseline characteristics. Values decrease predominantly with increases in HAQ score but also with increases in age and disease duration (32). Costs Drug costs, $/d Conventional 1.71 Fixed Methotrexate Biologic 47.36 Fixed Etanercept (31, 40) Other direct costs, $/d 12.17 Multivariate normal Value shown for patient with baseline characteristics. Values decrease with HAQ score, age, and disease duration (32). The 2004 rate was inflated to 2007 (40). Indirect costs Proportion working full time, % 67 Fixed (32) Average salary, $ 39 468 Fixed (32) Mortality Change in risk per point change in 1.33 Normal Based on a review by Chen et al, 2006 (41) HAQ score Change in risk for time receiving biologic 0.65 Beta From Jacobsson et al, 2007 (42). Rates are used to calculate annual probability of death by multiplication with life tables (43) and validated with standardized mortality ratios (44). Discount rate, % per year 3 Future costs and benefits are discounted as per guidelines (45). Time horizon Lifetime As per guidelines, we adopted a lifetime horizon but also report shorter time horizons. DMARD disease-modifying antirheumatic drug; EQ-5D EuroQol 5D; HAQ Health Assessment Questionnaire; NDB National Data Bank for Rheumatic Diseases. * As determined by HAQ score. The HAQ functional disability index ranges from 0 (no functional impairment) to 3 (maximum functional impairment). Maximum possible Larsen score, 200; maximum Genant-modified Sharp score, 202; maximum Sharp van der Heijde score, 448. Probability of excellent, good, moderate, or no treatment response 6 months after therapy, estimated by using a multivariate relationship based on age, sex, baseline HAQ score, current HAQ score, type of disease, education level, race, income, duration of disease, number of comorbid conditions, and type and number of previous received. The Appendix (available at www.annals.org) and Wailoo et al (31) contain parameter estimates. Initial nonsteroidal anti-inflammatory drugs, pain management, and low-dose glucocorticoids, followed by at 1 year. Methotrexate monotherapy. A tumor necrosis factor inhibitor in combination with methotrexate. ** Patients show minimal signs and symptoms of disease activity while receiving therapy (sometimes called remission in the rheumatology literature). Values depend on a multivariate relationship among disease duration, baseline HAQ score, age, sex, ethnicity, education, number of, number of comorbid conditions, income, and body weight. term radiographic progression by 33%, compared with delaying intervention for up to 5 years (49). Second, we included reductions in radiographic progression by type of treatment response in the model (36). We assumed no progression for patients who achieved an excellent or good response and reduced progression for moderate responders. We assumed that treatment would affect functional disability. The initial (first 6 months) improvement in HAQ score depends on the medication used, response, and disease duration for a given patient. We also assumed that the rate of HAQ score progression depended on response to therapy and on radiographic damage. The duration of treatment depended on the response to a particular therapy and was based on estimates from the NDB (15). A patient who withdraws switches to the next available treatment (31). This process is repeated for the patient s lifetime. Quality-Adjusted Life-Years Quality-adjusted life-years take both the quantity and quality of life into account by adjusting the duration of remaining life-years by quality or value. One year of perfect health equals 1.0 QALY, whereas 1 year of less-thanperfect health is less than 1.0 QALY; the magnitude of the reduction depends on the severity of the health state. We estimated the QALYs for various health states by using the mean health utility of each health state (45). We estimated EuroQol-5D (EQ-5D) health utility values (50) through their relationship with HAQ score, age, number of comorbid illnesses, and disease duration (15). We adjusted U.S. life tables (43) by standardized mortality rates for patients with RA (44) to estimate time of death. We adjusted risk for death for HAQ score and whether patients were receiving a biologic (42). Direct and Indirect Costs We allocated direct costs in 2 stages. First, we estimated all medical and social care costs, including joint replacements, as a function of HAQ score and other covariates. We estimated this relationship by drug type and included the cost of treating adverse events. We obtained unit costs for medications and outpatient and hospitalization costs from the Centers for Medicare & Medicaid Services and adjusted them to 2007 U.S. dollars by using the Consumer Price Index to be consistent with all costs in the model (40, 51). Next, we assigned the costs associated with the drugs in each strategy. The cost of therapy is zero for 616 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 www.annals.org

Cost Benefit Analysis for Treating Very RA Article patients in drug-free remission, a scenario we explored in sensitivity analyses. We incorporated indirect costs through the relationship between HAQ score and work capacity, as published elsewhere (52). We estimated the proportion of patients employed at baseline and their average salary on the basis of data from the NDB (32). In keeping with guidelines for conducting economic analyses, we discounted both future costs and benefits at a rate of 3% per year (50). Statistical Analysis We calculated ICERs by dividing the additional costs by the additional health benefit associated with one strategy relative to another (53). Because many data are based Table 2. Central Results and Sensitivity Analyses Scenario Description QALY Cost, $ ICER, $/QALY Pyramid Strategy Pyramid Strategy vs. Pyramid Strategy vs. Pyramid Strategy vs. Base case 14.7 15.0 14.8 131 890 133 340 196 003 4849 727 894 Dominated* 1. Lost productivity Productivity costs 14.7 15.0 14.8 177 001 172 006 224 569 Cost saving 540 054 Dominated* costs included included 2. Baseline HAQ a. 0.5 16.0 16.2 16.1 136 254 137 340 198 253 5117 494 365 Dominated* disability score b. 1.5 14.1 14.4 14.1 128 912 129 563 194 837 1930 Dominated* Dominated* 3. Mortality risk a. Only on the basis 14.5 14.9 14.7 128 985 130 853 195 114 5901 403 113 Dominated* reduction of HAQ score b. Only while 15.4 15.7 15.4 141 439 142 277 202 040 2802 988 004 Dominated* receiving biologics c. Lifelong effect of 15.1 15.4 15.4 136 674 138 135 201 785 4447 197 702 75 001 934 biologics 4. Response rates a. 10% improved 14.8 15.1 14.9 132 226 134 716 197 461 7332 392 039 Dominated* b. 10% worsened 14.6 15.0 14.7 131 055 133 442 196 457 6786 563 510 Dominated* c. 10% improved for 14.8 15.1 14.8 132 194 132 000 196 009 Cost saving 2 786 228 Dominated* only 5. Identification of Perfect identification 14.6 14.9 14.8 119 931 120 579 179 332 2135 479 798 Dominated* responders to biologics pretreatment 6. Time to a. Half the average 13.7 14.0 13.9 125 781 127 473 149 340 6539 149 013 Dominated* withdrawal for b. Double the 15.2 15.8 16.3 103 200 104 686 258 437 2673 141 474 284 056 each therapy type average 7. Number and a. Only 1 biologic 14.0 14.3 14.2 94 070 94 305 119 921 771 152 967 Dominated* placement of b. Only 2 biologics 14.4 14.7 14.5 115 708 115 745 161 902 119 458 987 Dominated* biologics in sequence c. Effective alternatives always available 16.0 16.3 17.2 174 247 180 173 313 362 16 878 116 163 157 350 8. Proportion of a. 0% 14.3 14.7 14.4 137 448 139 156 211 908 4687 637 582 Dominated* spontaneous b. 20% 15.4 15.7 15.6 124 566 126 615 177 576 7592 508 919 Dominated* remitters 9. Biologic annual a. 25% 14.7 15.0 14.8 83 970 82 428 93 781 Cost saving 111 389 Dominated* cost** b. 50% 14.7 15.0 14.8 99 943 99 398 127 855 Cost saving 316 891 Dominated* c. 125% 14.7 15.0 14.8 147 863 150 310 230 077 8185 933 395 Dominated* 10. induce Included 14.7 15.0 14.8 131 890 133 340 156 321 4849 277 377 Dominated* drug-free remission 11. Time horizon a. 5 y 3.7 3.8 3.9 17 311 16 626 69 548 Cost saving 235 750 443 086 b. 10 y 7.0 7.1 7.3 38 101 37 989 120 503 Cost saving 251 246 417 808 12. Discounting rate a. 0% 21.9 22.4 21.7 211 709 214 384 268 309 5438 Dominated* Dominated* for both costs b. 5% 11.7 11.9 11.8 98 331 99 309 162 703 4295 401 292 Dominated* and benefits 13. Best case Combination of scenarios 1, 3, 5, 6, and 9 15.4 16.05 17.8 86 177 85 085 169 804 Cost saving 35 071 46 903 DMARD disease-modifying antirheumatic drug; HAQ Health Assessment Questionnaire; ICER incremental cost-effectiveness ratio; QALY quality-adjusted life-year. * This strategy gives less benefit at more cost. The scenario takes the indirect costs associated with lost productivity into account. This strategy gives more benefit at less cost. Range, 0 3. Assumes that lower HAQ scores have a longer life expectancy, independent of treatment received. Only patients identified a priori as responders to biologics would receive these agents. ** Tumor necrosis factor inhibitors. Combines societal perspective (scenario 1), lifelong effect of biologics on mortality (scenario 3), perfect identification of responders to biologics (scenario 5), prolonged retention rate of antirheumatic agents (scenario 6), and reduced cost of biologics (scenario 9). www.annals.org 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 617

Article Cost Benefit Analysis for Treating Very RA HAQ Score Figure 2. Estimated progression of functional disability and radiographic joint damage over time for each strategy. Maximum Possible Joint Damage, % 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 0 5 10 15 20 25 Years 40 30 20 10 Pyramid strategy DMARD strategy biologic strategy 0 0 5 10 15 20 25 Years The evolution of functional disability and radiographic joint damage (erosion) after disease onset are displayed. The HAQ score ranges from 0 to 3 (maximum functional disability). Radiographic damage ranges from 0% (no joint damage) to 100% (total joint damage). DMARD disease-modifying antirheumatic drug; HAQ Health Assessment Questionnaire. on assumptions or are limited, we performed extensive sensitivity analyses. We assessed parameter uncertainty through a probabilistic sensitivity analysis by using a second-order Monte Carlo simulation to jointly examine the uncertainty in all model parameters (54). We varied individual assumptions and sources in a deterministic scenario analysis (Table 2) and repeated the probabilistic analyses after modifying the values of selected variables to assess their effect on the ICER (Table 1 and Appendix, available at www.annals.org). Role of the Funding Source Funding was provided by the Arthritis Foundation and an anonymous donor. The funding sources had no role in the conceptualization, design, conduct, or interpretation of the study or in the decision to submit the manuscript for publication. RESULTS The overall rate of progression of functional disability or joint damage is lower than in typical RCTs because our samples include patients with milder disease who would not be included in trials (Figure 2) (55). We had to make assumptions to extrapolate short-term data to longer time horizons, thereby increasing the uncertainty over time (Figure 2). Our model predicts that early therapy delays progressive joint damage, particularly for the first 10 years in the case of the early biologic strategy. Thus, the early DMARD and early biologic strategies delay HAQ score progression by 2 to 4 years (Figure 2); these delays influence both costs and QALYs by reducing use of resources, such as joint replacements and hospitalizations, and by preventing deterioration in health status. Table 2 shows mean lifetime costs and QALYs for the main strategies. Over a lifetime, the benefits of the early DMARD strategy are approximately 0.3 QALY higher than those of the pyramid strategy (15.0 [95% CI, 13.1 to 16.5] vs. 14.7 [CI, 12.8 to 16.3]). From a third-party payer perspective, the additional expenses are nearly offset by the costs saved through reduced hospitalizations (total lifetime costs, $133 340 [CI, $106 976 to $146,401] for the early DMARD strategy vs. $131 890 [CI, 103 569 to 145 432] for the pyramid strategy). This results in a favorable ICER of $4849 per QALY (CI, $0 to $16 354 per QALY). When lost productivity costs are taken into account, the cost savings increase and the early DMARD strategy dominates (provides greater benefit at less cost than) the pyramid strategy. The early biologic strategy offers the most benefit during the first 10 years (7.3 vs. 7.0 QALYs for the pyramid strategy). After that time, the model assumes that patients for whom 3 biologics have failed would revert to conventional, which are less effective in severe, established RA. Patients following this strategy therefore had less benefit over time than those on the other strategies, who began receiving biologics later in their disease course (Figure 2). Thus, we calculated the lifetime benefits of this strategy to be slightly less than for the early DMARD strategy. If we assume that an effective alternative biologic agent is always available, then the relative advantage is maintained for a lifetime (17.2 vs. 16.0 QALYs for the pyramid strategy) (Table 2, scenario 7c); however, this scenario increases the cost of the early biologic strategies by approximately $133 000 over a patient s lifetime, which results in an ICER of $157 350 per QALY (CI, $98 540 to $194 256 per QALY) compared with the early DMARD strategy. Table 2 also provides insight about when to use biologics in the sequence of antirheumatic therapy. The early DMARD strategy, which used biologics only after conventional DMARD failure, seemed to dominate early biologic therapy. It maximized the effectiveness of early and reserved biologics for those with greater treatment resistance of longer duration, for whom the incremental benefit is greater (33). In contrast, the early biologic strategy was more effective in the short term (yielding 0.2 to 0.3 more QALY than the other strategies in the first 10 years) 618 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 www.annals.org

Cost Benefit Analysis for Treating Very RA Article but required treating all patients with biologics, with attendant increased medication costs. We analyzed alternative scenarios to assess the robustness of the conclusions. We report only those that affected the results (Table 2). When we took a broader societal perspective and incorporated indirect costs associated with lost productivity, the incremental costs of the early treatment strategies were reduced (Table 2, scenario 1). Being able to determine responders without using biologic treatment (Table 2, scenario 5) would reduce the ICER of the early biologic strategy to $297 010 per QALY. Assuming that early biologic therapy can produce treatment-free remission in a significant proportion of patients (Table 2, scenario 10) would reduce the cost of the strategy by 20% and reduce the ICER to $244 310 per QALY. Other apparently influential parameters include a biologic s effect on mortality rate (Table 2, scenario 3d), drug retention (Table 2, scenario 6), the availability of effective alternatives after multiple failed biologics (Table 2, scenario 7c), and the cost of biologic therapy (Table 2, scenario 9). In the best-case scenario (Table 2, scenario 13), in which we combined scenarios 1, 3, 5, 6, and 9, the ICER of the early biologic strategy was as low as $46 900 compared with the early DMARD strategy. The Appendix (available at www.annals.org) shows results of the early DMARD and pyramid strategies without a biologic. DISCUSSION To our knowledge, this is the first analysis of the costeffectiveness of very early intervention in RA by using a model that examines radiologic progression with both reversible and irreversible dysfunction (35). We found that early intervention with a DMARD may be an attractive economic option, given the future costs that would probably be saved (for example, through reduced long-term joint damage and subsequent surgery). When we took a broader perspective and incorporated the indirect costs associated with lost productivity, the incremental costs of the early treatment strategies were further reduced. The additional drug costs for the early DMARD strategy are therefore offset, which leads to overall cost savings. The early biologic strategy offered the most benefit in certain scenarios, such as the assumption that patients who stop responding to a biologic will always have access to effective alternatives, which seems to be the current trend. Whether the additional drug acquisition costs are justified is debatable. Our findings are consistent with those of Chen and colleagues (41), who found that tumor necrosis factor inhibitors were more cost-effective as a last, not first, therapy. In early disease, patients with rapidly progressing disease cannot be reliably identified, and expensive therapy could therefore be given to patients who would have responded satisfactorily to conventional DMARD therapies. Our results should be interpreted carefully. A limitation of modeling cost-effectiveness in RA is that extrapolations of long-term benefits are often made with little or no long-term data. If RA treatments are to be compared with, for example, heart disease treatments meant to prolong life, the lifelong potential of RA therapies must be evaluated. Our analysis informs policymaking, identifies where uncertainty exists, and highlights areas for future research. We assumed that a patient s disease course cannot be predicted at outset. However, if treatments could be targeted to patients with severe progressive disease or to those likely to respond, more benefit would be realized. Some evidence suggests that these patients may be identifiable. Biomarkers for individual responsiveness to antirheumatic therapy, such as anticitrullinated peptide (56), the shared epitope (57), and magnetic resonance imaging (58), might be useful in this regard. In addition, if early biologic intervention could induce drug-free remission in a significant proportion of patients, then the reduced drug costs would substantially improve this strategy s ICER. However, solid evidence for this is still lacking (59). Finally, the current price of biologics is one of the most influential parameters in determining cost-effectiveness. A price reduction, from greater competition in the biologic market or the development of less expensive versions, would make the costeffectiveness arguments more justifiable. As noted, our results are subject to uncertainty that stems from imprecise or limited data. In particular, most RCTs in early RA have been performed in patients with disease of less than 2 years duration, whereas our time frame was treatment within 3 months of symptoms. Extrapolating efficacy data from populations with longer disease durations to patients with very early RA might underestimate the true benefit of early aggressive intervention. In addition, we derived many of the statistical relationships in this model from a cohort of patients with established RA, not very early RA. We performed extensive calibrations, informed by the literature (Appendix, available at www.annals.org), to evaluate the robustness of our conclusions. Our analysis did not include the additional costs for very early identification of RA. The societal perspective considered only the costs of productivity loss, and this was limited to absenteeism because of the lack of data. results of the BeSt (Behandel Strategieën [treatment strategies]) study (60) suggest that including more aspects of productivity and unpaid work could further improve the economic case for biologics, but the appropriateness of this is debatable for economic evaluations. Our analysis shows that therapeutic strategies involving early conventional or early biologics are preferred, but the additional costs of early biologics may not be justified for all patients. The most rational use of resources is prompt initiation of conventional DMARD therapy for patients with very early RA. From the University of Geneva, Geneva, Switzerland; University of British Columbia, Vancouver, British Columbia, Canada; University of Nebraska Medical Center, Omaha, Nebraska; National Data Bank for www.annals.org 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 619

Article Cost Benefit Analysis for Treating Very RA Rheumatic Diseases, Wichita, Kansas; and Brigham and Women s Hospital, Harvard Medical School, Boston, Massachusetts. Note: Dr. Finckh and Mr. Bansback contributed equally to this study. Acknowledgment: The authors thank Allan Wailoo, University of Sheffield, for giving us access to the initial model code. Grant Support: By an unrestricted research grant from the Arthritis Foundation and an anonymous donor. Potential Conflicts of Interest: Consultancies: A. Finckh (Wyeth, Essex, Sanofi-Aventis), A.H. Anis (Abbott, Wyeth). Honoraria: A.H. Anis (Wyeth). Grants pending: A.H. Anis (Abbott, Schering-Plough, Wyeth). Other: N. Bansback (OMERACT). Reproducible Research Statement: Study protocol: Not applicable. Statistical code: Available from Mr. Bansback (nbansback@cheos.ubc.ca). Data set: Certain portions of the analytic data set are available to approved persons through written agreements with the author or research sponsor; contact Mr. Bansback (e-mail, nbansback@cheos.ubc.ca). Requests for Single Reprints: Axel Finckh, MD, MS, Division of Rheumatology, Department of Internal Medicine, University Hospital of Geneva, 26 Avenue Beau-Séjour, CH-1211 Geneva 14, Switzerland. Current author addresses and author contributions are available at www.annals.org. References 1. Finckh A, Liang MH, van Herckenrode CM, de Pablo P. Long-term impact of early treatment on radiographic progression in rheumatoid arthritis: a metaanalysis. Arthritis Rheum. 2006;55:864-72. [PMID: 17139662] 2. Anderson JJ, Wells G, Verhoeven AC, Felson DT. Factors predicting response to treatment in rheumatoid arthritis: the importance of disease duration. Arthritis Rheum. 2000;43:22-9. [PMID: 10643696] 3. Landewé RB, Boers M, Verhoeven AC, Westhovens R, van de Laar MA, Markusse HM, et al. COBRA combination therapy in patients with early rheumatoid arthritis: long-term structural benefits of a brief intervention. Arthritis Rheum. 2002;46:347-56. [PMID: 11840436] 4. Quinn MA, Conaghan PG, O Connor PJ, Karim Z, Greenstein A, Brown A, et al. Very early treatment with infliximab in addition to methotrexate in early, poor-prognosis rheumatoid arthritis reduces magnetic resonance imaging evidence of synovitis and damage, with sustained benefit after infliximab withdrawal: results from a twelve-month randomized, double-blind, placebo-controlled trial. Arthritis Rheum. 2005;52:27-35. [PMID: 15641102] 5. Van der Bijl AE, Van der Kooij SM, Goekoop-Ruiterman YP, de Vries- Bouwstra JK, Breedveld FC, Allaart CF, et al. Persistent good clinical response after tapering and discontinuation of initial infliximab therapy in patients with early rheumatoid arthritis: 3 year results from the BEST trial [Abstract]. Presented at the Annual European Congress of Rheumatology, Amsterdam, the Netherlands, 21 24 June 2006. Ann Rheum Dis. 2006:P0180. 6. van Dongen H, van Aken J, Lard LR, Visser K, Ronday HK, Hulsmans HM, et al. Efficacy of methotrexate treatment in patients with probable rheumatoid arthritis: a double-blind, randomized, placebo-controlled trial. Arthritis Rheum. 2007;56:1424-32. [PMID: 17469099] 7. Donahue KE, Gartlehner G, Jonas DE, Lux LJ, Thieda P, Jonas BL, et al. Systematic review: comparative effectiveness and harms of disease-modifying medications for rheumatoid arthritis. Ann Intern Med. 2008;148:124-34. [PMID: 18025440] 8. Bathon JM, Martin RW, Fleischmann RM, Tesser JR, Schiff MH, Keystone EC, et al. A comparison of etanercept and methotrexate in patients with early rheumatoid arthritis. N Engl J Med. 2000;343:1586-93. [PMID: 11096165] 9. St. Clair EW, van der Heijde DM, Smolen JS, Maini RN, Bathon JM, Emery P, et al; Active-Controlled Study of Patients Receiving Infliximab for the Treatment of Rheumatoid Arthritis of Onset Study Group. Combination of infliximab and methotrexate therapy for early rheumatoid arthritis: a randomized, controlled trial. Arthritis Rheum. 2004;50:3432-43. [PMID: 15529377] 10. Breedveld FC, Weisman MH, Kavanaugh AF, Cohen SB, Pavelka K, van Vollenhoven R, et al. The PREMIER study: a multicenter, randomized, doubleblind clinical trial of combination therapy with adalimumab plus methotrexate versus methotrexate alone or adalimumab alone in patients with early, aggressive rheumatoid arthritis who had not had previous methotrexate treatment. Arthritis Rheum. 2006;54:26-37. [PMID: 16385520] 11. Goekoop-Ruiterman YP, de Vries-Bouwstra JK, Allaart CF, van Zeben D, Kerstens PJ, Hazes JM, et al. Clinical and radiographic outcomes of four different treatment strategies in patients with early rheumatoid arthritis (the BeSt study): a randomized, controlled trial. Arthritis Rheum. 2005;52:3381-90. [PMID: 16258899] 12. Keystone EC, Haraoui B, Bykerk VP. Role of infliximab in the treatment of early rheumatoid arthritis. Clin Exp Rheumatol. 2003;21:S200-2. [PMID: 14969078] 13. Ikeda K, Cox S, Emery P. Aspects of early arthritis. Biological therapy in early arthritis overtreatment or the way to go? Arthritis Res Ther. 2007;9:211. [PMID: 17540047] 14. Fleurence R, Spackman E. Cost-effectiveness of biologic agents for treatment of autoimmune disorders: structured review of the literature. J Rheumatol. 2006; 33:2124-31. [PMID: 17086602] 15. Brennan A, Bansback N, Nixon R, Madan J, Harrison M, Watson K, et al. Modelling the cost effectiveness of TNF-alpha antagonists in the management of rheumatoid arthritis: results from the British Society for Rheumatology Registry. Rheumatology (Oxford). 2007;46: 1345-54. [PMID: 17562686] 16. Hwang A, Gupta S, Liang MH. Identification of people with very early RA for optimal care: a public health challenge [Editorial]. Nat Clin Pract Rheumatol. 2008;4:114-5. [PMID: 18227830] 17. Detsky AS, Laupacis A. Relevance of cost-effectiveness analysis to clinicians and policy makers. JAMA. 2007;298:221-4. [PMID: 17622605] 18. Nell VP, Machold KP, Eberl G, Stamm TA, Uffmann M, Smolen JS. Benefit of very early referral and very early therapy with disease-modifying antirheumatic drugs in patients with early rheumatoid arthritis. Rheumatology (Oxford). 2004;43:906-14. [PMID: 15113999] 19. Aletaha D, Eberl G, Nell VP, Machold KP, Smolen JS. Practical progress in realisation of early diagnosis and treatment of patients with suspected rheumatoid arthritis: results from two matched questionnaires within three years. Ann Rheum Dis. 2002;61:630-4. [PMID: 12079906] 20. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum. 1988;31:315-24. [PMID: 3358796] 21. Chan KW, Felson DT, Yood RA, Walker AM. The lag time between onset of symptoms and diagnosis of rheumatoid arthritis. Arthritis Rheum. 1994;37: 814-20. [PMID: 8003053] 22. Kumar K, Daley E, Carruthers DM, Situnayake D, Gordon C, Grindulis K, et al. Delay in presentation to primary care physicians is the main reason why patients with rheumatoid arthritis are seen late by rheumatologists. Rheumatology (Oxford). 2007;46:1438-40. [PMID: 17578850] 23. van der Helm-van Mil AH, le Cessie S, van Dongen H, Breedveld FC, Toes RE, Huizinga TW. A prediction rule for disease outcome in patients with recentonset undifferentiated arthritis: how to guide individual treatment decisions. Arthritis Rheum. 2007;56:433-40. [PMID: 17265478] 24. van Aken J, van Dongen H, le Cessie S, Allaart CF, Breedveld FC, Huizinga TW. Comparison of long term outcome of patients with rheumatoid arthritis presenting with undifferentiated arthritis or with rheumatoid arthritis: an observational cohort study. Ann Rheum Dis. 2006; 65:20-5. [PMID: 15901632] 25. van Gaalen FA, Linn-Rasker SP, van Venrooij WJ, de Jong BA, Breedveld FC, Verweij CL, et al. Autoantibodies to cyclic citrullinated peptides predict progression to rheumatoid arthritis in patients with undifferentiated arthritis: a prospective cohort study. Arthritis Rheum. 2004;50:709-15. [PMID: 15022309] 26. van der Helm-van Mil AH, Detert J, le Cessie S, Filer A, Bastian H, Burmester GR, et al. Validation of a prediction rule for disease outcome in patients with recent-onset undifferentiated arthritis: moving toward individual- 620 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 www.annals.org

Cost Benefit Analysis for Treating Very RA Article ized treatment decision-making. Arthritis Rheum. 2008;58:2241-7. [PMID: 18668546] 27. Barton P, Bryan S, Robinson S. Modelling in the economic evaluation of health care: selecting the appropriate approach. J Health Serv Res Policy. 2004; 9:110-8. [PMID: 15099459] 28. Potter T, Mulherin D, Pugh M. intervention with disease-modifying therapy for rheumatoid arthritis: where do the delays occur? [Letter]. Rheumatology (Oxford). 2002;41:953-5; author reply 955. [PMID: 12154222] 29. Feldman DE, Bernatsky S, Haggerty J, Leffondré K, Tousignant P, Roy Y, et al. Delay in consultation with specialists for persons with suspected new-onset rheumatoid arthritis: a population-based study. Arthritis Rheum. 2007;57:1419-25. [PMID: 18050182] 30. Schmajuk G, Schneeweiss S, Katz JN, Weinblatt ME, Setoguchi S, Avorn J, et al. Treatment of older adult patients diagnosed with rheumatoid arthritis: improved but not optimal. Arthritis Rheum. 2007;57:928-34. [PMID: 17665462] 31. Wailoo AJ, Bansback N, Brennan A, Michaud K, Nixon RM, Wolfe F. Biologic drugs for rheumatoid arthritis in the Medicare program: a costeffectiveness analysis. Arthritis Rheum. 2008;58:939-46. [PMID: 18383356] 32. Wolfe F, Michaud K. A brief introduction to the National Data Bank for Rheumatic Diseases. Clin Exp Rheumatol. 2005;23:S168-71. [PMID: 16273802] 33. Nixon R, Bansback N, Brennan A. The efficacy of inhibiting tumour necrosis factor alpha and interleukin 1 in patients with rheumatoid arthritis: a meta-analysis and adjusted indirect comparisons. Rheumatology (Oxford). 2007; 46:1140-7. [PMID: 17478472] 34. Machold KP, Nell VP, Stamm TA, Eberl G, Steiner G, Smolen JS. The Austrian Arthritis Registry. Clin Exp Rheumatol. 2003;21:S113-7. [PMID: 14969061] 35. Aletaha D, Smolen J, Ward MM. Measuring function in rheumatoid arthritis: Identifying reversible and irreversible components. Arthritis Rheum. 2006;54: 2784-92. [PMID: 16947781] 36. Landewé R, van der Heijde D, Klareskog L, van Vollenhoven R, Fatenejad S. Disconnect between inflammation and joint destruction after treatment with etanercept plus methotrexate: results from the trial of etanercept and methotrexate with radiographic and patient outcomes. Arthritis Rheum. 2006;54:3119-25. [PMID: 17009230] 37. Wolfe F, Michaud K. Duration of Use of Anti-TNF Therapy in Rheumatoid Arthritis. Presented at the 2007 American College of Rheumatology Scientific Meeting, Boston, MA, 10 11 November 2007. Arthritis Rheum. 2007; S403. 38. Drossaers-Bakker KW, Kroon HM, Zwinderman AH, Breedveld FC, Hazes JM. Radiographic damage of large joints in long-term rheumatoid arthritis and its relation to function. Rheumatology (Oxford). 2000;39:998-1003. [PMID: 10986305] 39. Scott DL, Pugner K, Kaarela K, Doyle DV, Woolf A, Holmes J, et al. The links between joint damage and disability in rheumatoid arthritis. Rheumatology (Oxford). 2000;39:122-32. [PMID: 10725061] 40. U.S. Bureau of Labor Statistics. Consumer price index database. Washington, DC: U.S. Bureau of Labor Statistics; 2008. Accessed at www.bls.gov/cpi/ on 11 September 2009. 41. Chen YF, Jobanputra P, Barton P, Jowett S, Bryan S, Clark W, et al. A systematic review of the effectiveness of adalimumab, etanercept and infliximab for the treatment of rheumatoid arthritis in adults and an economic evaluation of their cost-effectiveness. Health Technol Assess. 2006;10:iii-iv, xi-xiii, 1-229. [PMID: 17049139] 42. Jacobsson LT, Turesson C, Nilsson JA, Petersson IF, Lindqvist E, Saxne T, et al. Treatment with TNF blockers and mortality risk in patients with rheumatoid arthritis. Ann Rheum Dis. 2007;66:670-5. [PMID: 17158824] 43. Arias E. United States Life Tables. National Vital Statistics Reports. Hyattsville, MD: National Center for Health Statistics; 2004. 44. Symmons D, Mathers C, Pfleger B. The Global Burden of Rheumatoid Arthritis in the Year 2000. Geneva: World Health Organization; 2000. 45. Gold M, Siegel J, Russell L, Weinstein M. Cost-Effectiveness in Health and Medicine New York: Oxford Univ Pr; 1996. 46. Fries JF, Spitz PW, Young DY. The dimensions of health outcomes: the health assessment questionnaire, disability and pain scales. J Rheumatol. 1982;9: 789-93. [PMID: 7175852] 47. Wolfe F. A reappraisal of HAQ disability in rheumatoid arthritis. Arthritis Rheum. 2000;43:2751-61. [PMID: 11145033] 48. van der Heijde D, Landewé R, Klareskog L, Rodríguez-Valverde V, Settas L, Pedersen R, et al. Presentation and analysis of data on radiographic outcome in clinical trials: experience from the TEMPO study. Arthritis Rheum. 2005;52: 49-60. [PMID: 15641062] 49. Finckh A, Simard JF, Gabay C, Guerne PA; SCQM physicians. Evidence for differential acquired drug resistance to anti-tumour necrosis factor agents in rheumatoid arthritis. Ann Rheum Dis. 2006;65:746-52. [PMID: 16339288] 50. Shaw JW, Johnson JA, Coons SJ. US valuation of the EQ-5D health states: development and testing of the D1 valuation model. Med Care. 2005;43:203-20. [PMID: 15725977] 51. Wailoo A, Brennan A, Bansback N, Nixon R, Wolfe F, Michaud K. Modeling the Cost Effectiveness of Etanercept, Adalimumab and Anakinra Compared to Infliximab in the Treatment of Patients With Rheumatoid Arthritis in the Medicare Program. Technology Assessment. Rockville, MD: Agency for Healthcare Research and Quality; 2006. Accessed at www.cms.hhs.gov/reports /downloads/wailoo.pdf on 17 September 2009. 52. Choi HK, Seeger JD, Kuntz KM. A cost effectiveness analysis of treatment options for methotrexate-naive rheumatoid arthritis. J Rheumatol. 2002;29: 1156-65. [PMID: 12064828] 53. Johannesson M, Weinstein MC. On the decision rules of cost-effectiveness analysis. J Health Econ. 1993;12:459-67. [PMID: 10131756] 54. Briggs AH. Handling uncertainty in cost-effectiveness models. Pharmacoeconomics. 2000;17:479-500. [PMID: 10977389] 55. Pincus T, Stein CM. Why randomized controlled clinical trials do not depict accurately long-term outcomes in rheumatoid arthritis: some explanations and suggestions for future studies. Clin Exp Rheumatol. 1997;15 Suppl 17:S27-38. [PMID: 9266130] 56. Nishimura K, Sugiyama D, Kogata Y, Tsuji G, Nakazawa T, Kawano S, et al. Meta-analysis: diagnostic accuracy of anti-cyclic citrullinated peptide antibody and rheumatoid factor for rheumatoid arthritis. Ann Intern Med. 2007; 146:797-808. [PMID: 17548411] 57. Gorman JD, Lum RF, Chen JJ, Suarez-Almazor ME, Thomson G, Criswell LA. Impact of shared epitope genotype and ethnicity on erosive disease: a metaanalysis of 3,240 rheumatoid arthritis patients. Arthritis Rheum. 2004;50:400-12. [PMID: 14872482] 58. Tamai M, Kawakami A, Uetani M, Takao S, Tanaka F, Fujikawa K, et al. Bone edema determined by magnetic resonance imaging reflects severe disease status in patients with early-stage rheumatoid arthritis. J Rheumatol. 2007;34: 2154-7. [PMID: 17918786] 59. van der Kooij SM, Goekoop-Ruiterman YP, de Vries-Bouwstra JK, Güler- Yüksel M, Zwinderman AH, Kerstens PJ, et al. Drug-free remission, functioning and radiographic damage after 4 years of response-driven treatment in patients with recent-onset rheumatoid arthritis. Ann Rheum Dis. 2009;68:914-21. [PMID: 18662933] 60. van den Hout WB, Goekoop-Ruiterman YP, Allaart CF, de Vries-Bouwstra JK, Hazes JM, Kerstens PJ, et al. Cost-utility analysis of treatment strategies in patients with recent-onset rheumatoid arthritis. Arthritis Rheum. 2009;61:291-9. [PMID: 19248130] www.annals.org 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 621

Annals of Internal Medicine Current Author Addresses: Dr. Finckh: Division of Rheumatology, University Hospital of Geneva, 26 Avenue Beau-Séjour, CH-1211 Geneva 14, Switzerland. Mr. Bansback, Dr. Anis, and Ms. Sizto: Centre for Health Evaluation and Outcome Sciences, University of British Columbia, St. Paul s Hospital, Vancouver, British Columbia, Canada. Dr. Marra: Faculty of Pharmaceutical Sciences, University of British Columbia, Vancouver, British Columbia, Canada. Dr. Michaud: 986270 Nebraska Medical Center, Omaha, NE 68198-6270. Dr. Lubin: General Practice, 101-777 Broadway West, Vancouver, British Columbia V5Z 4J7, Canada. Dr. White: Canadian Institute for the Relief of Pain and Disability, 204 916 West Broadway Avenue, Vancouver, British Columbia V5Z 1K7, Canada. Dr. Liang: Division of Rheumatology, Immunology, and Allergy, Department of Medicine, Harvard Medical School, Brigham and Women s Hospital, Francis Street, PBB-3, Boston, MA 02115. Author Contributions: Conception and design: A. Finckh, N. Bansback, C.A. Marra, A.H. Anis, S. Lubin, M. White, M.H. Liang. Analysis and interpretation of the data: N. Bansback, C.A. Marra, K. Michaud, M. White, S. Sizto, M.H. Liang. Drafting of the article: A. Finckh, N. Bansback, A.H. Anis, K. Michaud, S. Sizto, M.H. Liang. Critical revision of the article for important intellectual content: A. Finckh, N. Bansback, C.A. Marra, A.H. Anis, M. White, M.H. Liang. Final approval of the article: A. Finckh, N. Bansback, C.A. Marra, A.H. Anis, S. Lubin, M. White, M.H. Liang. Statistical expertise: A. Finckh, N. Bansback, A.H. Anis. Obtaining of funding: A.H. Anis, M.H. Liang. Administrative, technical, or logistic support: A. Finckh, A.H. Anis, M. White, M.H. Liang. Collection and assembly of data: A. Finckh, N. Bansback, A.H. Anis, K. Michaud, M. White, S. Sizto, M.H. Liang. W-198 3 November 2009 Annals of Internal Medicine Volume 151 Number 9 www.annals.org