Review of Eight Pharmacoeconomic Studies of the Value of Biologic DMARDs Adalimumab, Etanercept, and Infliximab) in the Management of Rheumatoid Arthritis

BACKGROUND: Treatment options for the management of rheumatoid arthritis (RA) have expanded from the traditional disease-modifying antirheumatic drugs (DMARDs) to include the biologic DMARDs that inhibit tumor necrosis factoralpha (TNF-a). OBJECTIVES: To assess the medical literature for studies of the economic value of biologic DMARDs, specifically the 3 TNF-a inhibitors (adalimumab, etanercept, and infliximab) used for the management of RA, compared with the traditional DMARDs such as sulfasalazine, antimalarials, penicillamine, gold, methotrexate, azathioprine, leflunomide, and cyclophosphamide. METHODS: A comprehensive search of the MEDLINE and HealthSTAR databases was conducted to identify cost-efficacy, cost-effectiveness, or cost-utility studies published in the English language (from 1966 through November 2004). The search terms and/or MeSH (medical subject headings) titles were cost-benefit analysis, rheumatoid arthritis, antirheumatic agents, antineoplastic and immunosuppressive agents. Studies were critically reviewed and quality was assessed using the Quality of Health Economic Studies instrument. Most studies evaluated the use of biologics among RA patients resistant to DMARDs. Studies were assessed with regard to comparators evaluated, measures of efficacy, perspectives, model duration, treatment duration, and discount rate. RESULTS: From 180 titles identified, 155 were excluded for the following reasons: 89 because they did not consider the drugs of interest, 15 because the population was not RA, 19 because of having the wrong drugs and population, 22 because they were review articles, and 10 because they were general articles. Twenty five abstracts were accepted for further review. Of these, 13 abstracts were subsequently selected for full-text review. One of the authors identified a study not indexed in MEDLINE. Ultimately, 2 cost-effectiveness and 6 cost-utility studies were selected for this critical review. One study over 6 months reported that triple therapy with DMARDs (methotrexate-hydroxychloroquine-sulfasalazine) was cost effective for methotrexate-resistant patients, which is consistent with American College of Rheumatology (ACR) guidelines that support the use of triple therapy prior to biologics. The incremental cost-effectiveness ratio (ICER) was $1,500 per patient to achieve an ACR20 response for this triple therapy compared with no second-line agent. Overall, biologic therapies cost considerably more than traditional DMARDs but produced more quality-adjusted life-years (QALYs). Despite differences in design and assumptions, published economic models consistently reported ICERs less than$50,000 per QALY gained for biologics compared with traditional DMARDs, although ICERs of greater than$100,000 were reported from sensitivity analyses. CONCLUSIONS: Clinical guidelines currently recommend the use of biologics as step therapy after failure of traditional DMARDs. Reported ICERs comparing biologics with traditional DMARDs are within a range that is comparable with other accepted medical interventions. The worth of the additional expenditure will ultimately be judged by formulary and policy decision makers because no maximum cost has been defined. Models can be used to inform decision makers, but they must be interpreted and applied carefully. More research is also needed to differentiate the relative economic value of the various biologic agents by therapeutic indication.

traditional DMARDs, [5][6][7][8] but they are associated with greater costs (including costs of drugs and of health resource utilization). These costs, when accumulated over the duration of the condition, are of interest to potential payers.
In recent years, a number of evaluations have assessed the economic value of biologics for the management of RA. Several review papers have been published based on this body of literature, which either focused on comparing the underlying methodologies across studies or provided a review of a single biologic. [9][10][11][12] In contrast, the present review aims to provide decision makers with the results of research performed to determine the potential economic value of biologic DMARDs (i.e., TNF-α inhibitors) and to highlight special considerations when interpreting results for formulary decisions. Specifically, the objective of this article is to provide a comprehensive review of the literature on cost-effectiveness analyses (CEAs) and cost-

TNF-α Inhibitor FDA-Approved Indication
Infliximab* Rheumatoid arthritis: • Infliximab, in combination with methotrexate, is indicated for reducing signs and symptoms, inhibiting the progression of structural damage, and improving physical function in patients with moderately to severely active rheumatoid arthritis. Crohn's disease: • Infliximab is indicated for reducing signs and symptoms and inducing and maintaining clinical remission in adult and pediatric patients with moderately to severely active Crohn' s disease who have had an inadequate response to conventional therapy. • Infliximab is indicated for reducing the number of draining enterocutaneous and rectovaginal fistulas and maintaining fistula closure in adult patients with fistulizing Crohn' s disease. Ankylosing spondylitis: • Infliximab is indicated for reducing signs and symptoms in patients with active ankylosing spondylitis. Psoriatic arthritis: • Infliximab is indicated for reducing signs and symptoms of active arthritis in patients with psoriatic arthritis. Ulcerative colitis: • Infliximab is indicated for reducing signs and symptoms, achieving clinical remission and mucosal healing, and eliminating corticosteroid use in patients with moderately to severely active ulcerative colitis who have had an inadequate response to conventional therapy.

Etanercept † Rheumatoid arthritis:
• Etanercept is indicated for reducing signs and symptoms, inducing major clinical response, inhibiting the progression of structural damage, and improving physical function in patients with moderately to severely active rheumatoid arthritis. Etanercept can be initiated in combination with methotrexate or used alone. Polyarticular-course juvenile rheumatoid arthritis: • Etanercept is indicated for reducing signs and symptoms of moderately to severely active polyarticular-course juvenile rheumatoid arthritis in patients who have had an inadequate response to one or more DMARDs. Psoriatic arthritis: • Etanercept is indicated for reducing signs and symptoms, inhibiting the progression of structural damage of active arthritis, and improving physical function in patients with psoriatic arthritis. • Etanercept can be used in combination with methotrexate in patients who do not respond adequately to methotrexate alone. Ankylosing spondylitis: • Etanercept is indicated for reducing signs and symptoms in patients with active ankylosing spondylitis. Plaque psoriasis: • Etanercept is indicated for the treatment of adult patients (18 years or older) with chronic moderate-to-severe plaque psoriasis who are candidates for systemic therapy or phototherapy.

Adalimumab ‡ Rheumatoid arthritis:
• Adalimumab is indicated for reducing signs and symptoms, including major clinical response, inhibiting the progression of structural damage and improving physical function in adult patients with moderately to severely active rheumatoid arthritis. Adalimumab can be used alone or in combination with methotrexate or other DMARDs. Psoriatic arthritis: • Adalimumab is indicated for reducing signs and symptoms of active arthritis in patients with psoriatic arthritis. Adalimumab can be used alone or in combination with DMARDs. Ankylosing spondylitis: • Adalimumab is indicated for reducing signs and symptoms in patients with active ankylosing spondylitis. utility analyses (CUAs) of biologic DMARD treatments for RA, specifically for the 3 TNF-α inhibitors (adalimumab, etanercept, and infliximab).
The TNF-α inhibitors adalimumab, etanercept, and infliximab (the latter only in combination with methotrexate) are recommended as options for the treatment of adults who have both of the following characteristics: (1) continuing clinically active and severe RA as measured by disease activity score (DAS28) >5.1 (i.e., highly active disease)-disease activity should be measured at 2 time points, 1 month apart, confirming ongoing active disease; and (2) (Chiou). Data were collected on the comparators studied, patient characteristics, data sources, model assumptions, costs, effectiveness, and incremental cost-effectiveness ratios (ICERs). These studies reported costs from different years and in different currencies. To reduce the variation, costs were reported in 2 ways: (1) expressed as the value documented in the year when the analysis was conducted and (2) converted to 2004 U.S. dollars (USD) using the medical care component of the Consumer Price Index (CPI) for studies done in prior years. 14 For studies that expressed costs in currencies other than USD, a currency exchange rate was applied to convert their values into USD, and then the CPI was applied to adjust costs to their 2004 values. In this article, costs represent the values in the base year during which the authors conducted their models and/or analyses. Where indicated, adjusted costs represent values in 2004 USD.

Comparison of Study Characteristics
Because the value of a reported ICER depends on which reference comparator was chosen, large differences could be observed across studies that employed different methods of calculation. For this review, 2 methods were used to standardize the reporting of the ICERs. In the first method, comparators were rank ordered from the least to the most costly. Alternatives that were both more costly and less effective than another option (i.e., dominated) were eliminated from consideration. ICERs were calculated and reported among the remaining alternatives. Alternatively, the ICERs were calculated using a common reference comparator; in most instances, the comparator was methotrexate.
The quality of each study was assessed using the Quality of Health Economic Studies (QHES) instrument. 15,16 This instrument has 16 criteria that cover areas of methodology, valid and transparent results, and comprehensive reporting of the results. 15 Each criterion has a weighted point value: a maximum total score of 100 is possible, and a higher score implies better quality. A study with a score >75 can be considered of "good" quality. Furthermore, having a score that represents the quality of a study could be useful to identify studies that should receive more attention and be given greater weight in the decisionmaking process. Greater familiarity and application of the QHES instrument could facilitate systematic evaluation of costeffectiveness literature.

ss Results
From a total of 180 titles identified, 155 were excluded for the following reasons: 89 because they did not consider the drugs of interest, 15 because the population was not RA, 19 because they had the wrong drugs and population, 22 because they were review articles, and 10 because they were general articles. Twenty-five abstracts were accepted for further review. Of these, 13 were subsequently selected for full-text review. Three of the 13 were excluded because the drugs of interest were not included. Another 3 of the 13 were excluded because they were review articles. One of the authors identified a study not indexed in MEDLINE. Ultimately, 2 cost-effectiveness and 6 cost-utility studies were selected for this critical review. Two were CEAs, which defined effectiveness based on ACR criteria. ACR 20 was defined as ≥20% improvement in tender and swollen joint counts and ≥20% improvement in 3 of 5 other core measures: patient' s global assessment, physician' s global assessment, physical disability score, acute-phase reactant value, and patient' s assessment of pain. 17 The remaining studies were CUAs, which defined effectiveness as quality-adjusted life-years (QALYs). Table 2 shows distinct variation across studies in terms of comparators evaluated, perspectives, model duration, treatment duration, and discount rate.
Using the QHES instrument, all studies achieved scores of ≥78, and scores ranged from 78 to 92 (Table 3). Studies with lower scores tended to evaluate RA over a time period of <1 year, did not discuss direction and magnitude of potential biases, and/or did not adequately present study limitations.  1 Was the study objective presented in a clear, specific, and measurable manner? (7 points) 2 Were the perspective of the analysis (societal, third-party payer, etc.) and reasons for its selection stated? (4 points) 3 Were variable estimates used in the analysis from the best available source (i.e., Randomized Control Trial-Best, Expert Opinion-Worst)? (8 points) 4 If estimates came from a subgroup analysis, were the groups pre-specified at the beginning of the study? (1 point) 5 Was uncertainty handled by: 1) statistical analysis to address random events; 2) sensitivity analysis to cover a range of assumptions? (9 points) 6 Was incremental analysis performed between alternatives for resources and costs? (6 points) 7 Was the methodology for data abstraction (including value health states and other benefits) stated? (5 points) 8 Did the analytic horizon allow time for all relevant and important outcomes? Were benefits and costs that went beyond one year discounted (3%-5%) and justification given for the discount rate? (7 points) 9 Was the measurement of costs appropriate and the methodology for the estimation of quantities and unit costs clearly described? (8 points) 10 Were the primary outcome measure(s) for the economic evaluation clearly stated and were the major short term, long term and negative outcomes included? (6 points) 11 Were the health outcomes measures/scales valid and reliable? If previously tested valid and reliable measures were not available, was justification given for the measures/scales used? (7 points) 12 Were the economic model (including structure), study methods and analysis, and the components of the numerator and denominator displayed in a clear transparent manner? (8 points) 13 Were the choice of economic model, main assumptions and limitations of the study stated and justified? (7 points) 14 Did the author(s) explicitly discuss direction and magnitude of potential biases? (6 points) 15 Were the conclusions/recommendations of the study justified and based on the study results? (8 points) 16 Was there a statement disclosing the source of funding for the study? (3 points) as either the ACR20 or the ACR70 weighted response, methotrexate was the lowest-cost option and etanercept was the highestcost option. The least effective option was no second-line agent at 0.27, and the most effective option was etanercept at 0.68. Compared with methotrexate, etanercept was associated with an ICER of $40,300 per patient with an ACR20 response ($49,900, 2004 USD) over a 6-month period (Table 5). This is interpreted as the additional cost per patient to achieve an ACR20 response. In 1-way sensitivity analyses, the ICER per patient achieving ACR20 improvement for etanercept was greater than $39,000 unless the cost of etanercept was reduced or the probability of achieving ACR20 response was increased. When the baseline cost of etanercept ($6,600) was reduced by 25% and 50%, the ICERs compared with methotrexate were $28,400 and $15,000, respectively, per ACR20. When the probability of achieving ACR20 (81%) was increased by 20%, the ICER was $17,700 compared with methotrexate. Among methotrexate-resistant patients, Choi et al. analyzed  19 Efficacy data were based on 3 double-blind randomized controlled trials (RCTs) and 1 open-label trial. ACR response data were collected from the source clinical trials and used to estimate the probabilities of achieving an ACR response for each of the comparators. The authors assumed that patient characteristics in the source trials were similar based on similar RA duration and Health Assessment Questionnaire (HAQ) disability score. HAQ assesses arthritis-related functional disability in activities such as dressing, arising, eating, walking, hygiene, and reaching and gripping. 20 The HAQ score ranges from 0 to 3; a higher score indicates greater disability. Also, combination therapies were assumed to be associated with no more adverse effects than methotrexate monotherapy, which was suggested by findings from individual trials. Defining ACR20 as the effectiveness measure resulted in the "no secondline agent" option being the lowest cost and least cost-effective option, and methotrexate plus etanercept being the highest-cost and most cost-effective option, over the 6-month model duration (Table 5). When compared with methotrexate, the triple therapy option (methotrexate-hydroxychloroquinesulfasalazine) resulted in an ICER of $1,500 per patient with ACR20 response over a 6-month period. The ACR guidelines support the use of triple therapy prior to biologics, and this study supported the cost-effectiveness of this strategy. Etanercept monotherapy was calculated to have an ICER of $10,700 per patient with ACR20 response ($13,220, 2004 USD) when compared with methotrexate. The ICER for the methotrexate-plus-etanercept option was calculated to be $10,900 per patient with ACR20 response ($13,600, 2004 USD).

Cost-Utility Studies: Cost per QALY Gained
Six studies used QALYs as the effectiveness measure. Of these, 2 were conducted in the United States and 4 were in other countries. These economic models were based on studies that evaluated RA patients who failed at least 1 DMARD and/or who were methotrexate-resistant. Because of inadequate response to previous trial(s) of DMARDs, patients entering these models were considered eligible for biologics.

U.S. Studies
Wong et al. (Table 6)    • Did not include non-treatment-related adverse events, potential improvement in long-term clinical outcomes, or indirect costs. • Effectiveness is measured at 6 months and 12 months.

Characteristics of Cost-Utility Studies
• Where 12-month effectiveness rates were not available, 6-month and 12-month effectiveness rates were assumed to be equivalent.  To estimate the costutility of infliximab (initial treatment at weeks 0, 2, and 6, then given at either 3 mg/kg or 10 mg/kg dose every 4 or 8 weeks) plus methotrexate (≥12.5 mg QW) compared with methotrexate alone (≥12.5 mg QW) in inadequately controlled RA • Treatment was stopped after 1 or 2 years, when no further clinical data were available, and no further treatment costs and effects were therefore assumed. • NSAID usage was not included, as most patients used them and usage did not differ significantly between states. • Clinical benefits diminished over time and not immediately at discontinuation of infliximab. • Cycle length of 1 year

Cost of hospitalization:
Based on the number of inpatient days in different wards and ward-specific daily costs Surgical intervention: Calculated from the type of intervention and its duration multiplied by the cost per minute of operating theatre use Outpatient care: Based on the number of visits to different health care professionals Drug cost: Calculated from the number of months of use of each drug, associated with the cost of standard drug monitoring protocols in place in the rheumatology departments participating. Unit costs were taken from hospital accounting data and official price lists from National Health Service (U.K.) and University Hospital Lund (Sweden). Indirect costs: Calculated using the human capital approach, in which an individual' s productivity is valued at the market price To evaluate costs, benefits, and costeffectiveness of etanercept or infliximab treatment over 1-year period compared with no biologic • Comparator represented a group with costs and benefits that were established from baseline and were assumed to remain the same throughout the year. That is, comparison with another RA agent was not conducted. • Improvement in utility occurred after 3 months of treatment (base case).

Structured interview:
Obtained resource consumption and work capacity data for the year before treatment and the first anti-TNF year Indirect costs: Estimated by human capital method using average annual gross salary; sick days and loss of productivity were included Sensitivity analyses revealed that the cost of biologics and probabilities for achieving ACR response were the main drivers of incremental cost-effectiveness ratios. Because the costeffectiveness of biologics relative to nonbiologic agents was not compared, the findings from this study cannot be directly compared with those of other studies.

Swedish and United Kingdom Studies
Four studies examined the cost-utility of biologics from a non-U.S. perspective. Kobelt et al. presented results from both the Swedish and U.K. perspectives. 23 Brennan et al. provided only a U.K. perspective. 24 Kobelt et al. (2003) estimated the cost-utility of infliximab plus methotrexate compared with methotrexate alone in RA patients not adequately controlled with traditional DMARDs (Table 6). 23 A Markov model was constructed with health states defined as functional disability levels (as measured by HAQ scores), and a death state. The model distributed patients into different health states based on whether their HAQ scores have improved, remained stable, or worsened, or if the patient died during the cycle. Although the duration of the model was 10 years, the treatment effects of biologics beyond 2 years were not modeled because long-term clinical data were not available.
First-year efficacy data were taken from the ATTRACT (Anti-TNF Trial in Rheumatoid Arthritis with Concomitant Therapy) study, and data for years 2 through 10 were based on epidemiological observation of HAQ disability profile from the Swedish Lund Cohort study. The combination of infliximab plus methotrexate (Table 7) was both more costly and more effective compared with methotrexate alone, but the gains in QALYs were associated with a favorable cost-effectiveness profile. Results were qualitatively similar in the 1-year and 2-year analyses of biologic treatment.
In the same study, a separate analysis from the U.K. perspective was presented (Table 6). 23 This model mirrored the Swedish model except that the long-term data beyond year 2 were based on a cohort from the Early RA Study 27 in the United Kingdom. The combination of infliximab plus methotrexate was also found to be more expensive and more efficacious compared with methotrexate alone ( Table 7). The ICER was calculated to be £21,600 (British pounds) per QALY gained ($48,710, 2004 USD). Kobelt et al. (2004) evaluated the cost-effectiveness of etanercept or infliximab compared with routine clinical practice (i.e., without anti-TNF) for the treatment of patients with RA in Sweden (Table 6). 25 Unlike other studies that were model-based analyses, this study collected actual data on direct and indirect costs, health-related quality of life (HRQoL), and HAQ scores from patients who were either resistant or intolerant of at least 2 traditional DMARDs including methotrexate. The authors concluded that the use of etanercept or infliximab in this population was cost effective because the ICER was below the generally accepted 50,000 EUR (Euros) per additional QALY gain threshold. The following yielded ICERs that did not exceed that threshold: sensitivity analyses conducted on the direct cost only, utility improvement after 6 weeks (instead of 3 months), and linear improvement in utility over 1 year. The ICERs surpassed this threshold only when an intent-to-treat analysis (including all dropouts) was conducted or when patients with low disability (HAQ score <1.6) at baseline were considered (Table 7).¤ Bansback et al. conducted a lifetime CUA comparing adalimumab, etanercept, and infliximab as monotherapy and as combination therapy with methotrexate, compared with traditional DMARDs, from the perspective of a policy decision maker (Table 6). 26 A hypothetical population of 10,000 patients who had failed to respond to a traditional DMARD and who were eligible for biologics entered the model. After failure with biologics, 3 other DMARDs were tried. Two versions of the model were created and analyzed based on patients achieving an ACR20 response and an ACR50 response, but these responses were translated to DAS28 response criteria. The results for adalimumab were based on the pooled results of 2 trials. HRQoL and costs were modeled as a function of the HAQ disability index. In the ACR50 version of the analysis, single and combination therapies with biologics were more costly but produced more QALYs compared with DMARDs. It is worth noting that the estimated QALY of <3 from this model was low considering that a lifetime analysis was conducted. In the basecase results, the ICERs were comparable across all biologics, but adalimumab plus methotrexate was the lowest at 34,922 EUR Patients who entered the model had failed at least 2 DMARDs. 24 The model compared patients who continued on a protocol of either a series of traditional DMARDs or etanercept monotherapy (followed by the above DMARD series in case of adverse effects or lack of efficacy). Using etanercept as a monotherapy after failures with traditional DMARDs was more costly (Table 7) but more efficacious, and was associated with a favorable costeffectiveness profile (£16,330 per QALY gained [$29,433 in 2004 USD] compared with treatment using a series of DMARDs). The results were relatively insensitive to changes in key parameters, with the ICERs ranging between £14,000 and £21,000 per QALY gained ($25,233 and $37,850, 2004 USD, respectively).

ss Discussion
The introduction of biologic DMARDs for the management of RA poses several challenges for health care decision makers, especially in an era when drug expenditures continue to rise and cost containment is common. First, at an average annual drug cost of $17,000 to $18,000, biologic DMARDs are much more costly than traditional DMARDs. The annual cost of methotrexate therapy is approximately $200 (assuming 7.5 mg per week). Moreover, the studies reviewed have consistently shown that the additional costs of biologics are not completely offset by preventing future disability; hence, the clinical benefits of biologic therapy are likely to come with additional costs. These facts cause payers to be concerned about the value for money obtained from the use of biologic therapy. In the process of trying to determine formulary placement of these expensive specialty drugs, benefit designers also need to consider member access and cost sharing.
Second, the demand for biologics from physicians and patients may increase with more data from RCTs indicating that biologics improve ACR response and decrease disability. Managed care organizations must balance the evidence on safety, efficacy, effectiveness, and cost to assess the economic value of biologic therapy. Well-conducted CEAs are potentially helpful here.
The studies reviewed here suggest that the additional benefits of biologics after failing traditional DMARDs may be worth the additional cost compared with DMARD continuation, based on the commonly cited thresholds used in the different countries. These thresholds, also referred to as ICERs, represent the additional amount of money that payers will spend to gain 1 additional QALY (i.e., a year of perfect health) compared with the current gold standard or best therapy option. While no threshold formally exists in the United States, historically an ICER of <$50,000 per QALY gained has been cited as a good value for the additional spending. In other words, the drug can be considered cost effective if the ICER is ≤$50,000. However, an ICER approaching $100,000 per QALY gained has also been used to justify additional drug spending. 28 The ICERs reported in the studies in our review are within or below this range, even after adjusting for inflation and differences in currencies. In the United States, the adjusted ICER from Wong et al. 21 was $11,670 per QALY gained for infliximab plus methotrexate compared with methotrexate alone (Table 7). In the United Kingdom, these adjusted ICERs ranged from $29,433 (for etanercept compared with traditional DMARDs) to $48,710 (for infliximab plus methotrexate compared with methotrexate alone) ( Table 7). Studies from the Swedish setting reported adjusted ICERs ranging from $3,500 (for infliximab plus methotrexate compared with methotrexate alone) to $85,395 (for adalimumab compared with DMARDs). Despite the dissimilarity in their methodologies, these studies consistently reported ICERs within the range where payers may accept additional spending for these agents.
Pharmacoeconomic models may aid decision making in several ways. First, primary collection of data on costs and utilization in an RCT is often impractical; models can overcome this limitation because they can synthesize information from disparate sources. Second, a drug is often compared with placebo in an RCT, but decision makers need to know how the drug compares with standard therapy. Third, models can be used to project the long-term costs and consequences for a chronic condition such as RA in the absence of actual data from RCTs. Lastly, the results from a well-designed model can be presented in a useful metric such as cost per QALY gained, so that therapies can be compared within RA and across other conditions. Policy makers using these models to make decisions will need to inspect how the study was conducted, to verify the face-validity of key assumptions and to determine whether the model was framed in a way that answered the relevant questions. Several questions need to be addressed before applying the results to the payer' s population.
First, are the metrics reported from these studies useful? Modeling the effective measure as a function of ACR response is practical because this outcome is often reported in clinical trials. For example, the 2 studies by Choi et al. reported the incremental cost per additional patient with ACR response. 18,19 Response rate is useful when decision makers are only concerned with comparing relative efficacies of different agents. However, this metric has limited application for formulary decision making in the context of cost-effectiveness because the results cannot be easily compared with those of other economic Group' s model were that (1) using anti-TNF-α as a first-line treatment is not cost effective (ICER >£100,000 per QALY gained) and (2) the assumptions relating to HAQ progression have a significant impact on the cost-effectiveness estimates. If the model assumed no progression of disease while responding to treatment (i.e., optimistic scenario), the ICERs (versus no biologic treatment) were £30,000 per QALY gained for etanercept, £58,000 per QALY gained for adalimumab, and £55,000 per QALY gained for infliximab. If disease progression was assumed to occur during biologic treatment (i.e., the conservative basecase scenario), the ICERs (versus no biologic treatment) were £88,300 per QALY gained for etanercept.
After considering all the economic evidence, the NICE Appraisal Committee concluded that "some patients with severe active disease who have failed to respond to at least 2 trials of DMARDs could be identified and managed cost effectively using 1 of the 3 TNF-α inhibitors." 13 Therefore, the committee did not recommend use of a TNF-α inhibitor in early stages of RA. Furthermore, treatment should be discontinued if response is not maintained, defined as clinical deterioration (i.e., increase of DAS28 score by more than 0.6) at consecutive assessments. The findings from the studies in our review support these preliminary recommendations of the NICE Appraisal committee (i.e., patients should have failed traditional DMARDs prior to use of a TNF-α inhibitor.) In addition, the committee appeared to support a conservative approach that analytic models should show the loss of clinical benefits at treatment withdrawal (i.e., recurrence of disability) rather than assume that benefits can be extended beyond discontinuation.
In addition to our qualitative review of these studies, we found that reporting scores from the QHES instrument can be useful to decision makers, particularly those with limited experience with reviewing pharmacoeconomic data, as a way to differentiate one study from another. Ofman and colleagues reported that "experts" in health economics considered the QHES instrument as moderately valuable, but 54% would still recommend it to others. 16 On a scale of 1 ("not valuable at all") to 5 ("extremely valuable"), these experts rated the tool with a mean of 3.6 (±1.0). More importantly, Ofman and colleagues also reported that among those not considering themselves as experts in evaluating health economic analyses, 76% responded that they would use this tool. In settings where economic evaluations are considered when decisions about resource allocation are made, a tool such as the QHES can be useful.

Limitations
Four of the 8 economic evaluations reviewed were sponsored by manufacturers of TNF-α inhibitors, and 1 of the 8 studies was sponsored by a manufacturer and not indexed by MED-LINE. Bell and colleagues noted that published cost-effectiveness studies tend to report favorable incremental cost-effectiveness ratios. 30 Furthermore, Bell et al. found studies funded by industry to be more likely to report ratios below $20,000, $50,000, and $100,000 per QALY gained. However, studies of higher methodological quality and those conducted in Europe or the United States were less likely to report ratios below $20,000 per QALY gained. These observations suggest that decision makers need to consider study sponsorship and inspect such studies more critically for any potential biases. However, industry sponsorship does not necessarily discredit the findings from such studies.
These models relied on efficacy data from source clinical studies in which the patients had failed at least 1 traditional DMARD. Only 3 of the 8 models defined failure to respond to traditional DMARDs as failing at least 2 traditional DMARDs, of which 1 has to be methotrexate ( Table 2). [24][25][26] This definition of failure is consistent with the recommendation from the BSR guidelines to determine when patients become eligible for biologic therapies. 4 The 3 studies that evaluated cost-effectiveness of infliximab (in combination with methotrexate) focused on patients with inadequate response as methotrexate resistant but not necessarily having failed 2 DMARDs. 19,21,23 Arguably, these patients may respond to other less costly traditional DMARD before considering biologic therapies.
A gap exists between clinical practice guidelines and formal indications approved by the U.S. Food and Drug Administration. According to the current prescribing information (Table 1), etanercept and adalimumab can be initiated in combination with methotrexate or used alone. However, patients are rarely prescribed biologics without having tried at least 1 and usually 2 or more traditional DMARDs in the real-world practice setting The cost-effectiveness literature assessed the use of biologic therapies across a range of clinical circumstances from multiple treatment failures with DMARDs to 1 study that involved DMARD-naïve patients. The resulting economic outcomes for these diverse clinical scenarios are consistent with the amount paid for other therapeutic interventions. In a circumstance where only specific clinical situations meet cost-effectiveness guidelines, then a narrowed therapeutic use could be defined. However, we found no such limitation in the pharmacoeconomic literature, and the cost-effectiveness literature has not yet addressed step therapy with DMARDs followed by biologics.

ss Conclusions
Biologic therapies are more costly compared with traditional DMARDs but produce more QALYs. Despite differences in design and assumptions, published economic models consistently reported ICERs <$50,000 per QALY gained for biologics compared with traditional DMARDs when used among RA patients who have become resistant to DMARDs, although sensitivity analyses reported ICERs of >$100,000. This implies that the value of biologics is comparable with that of other wellaccepted medical interventions. Nonetheless, the formulary and policy decision makers will ultimately have to judge whether the additional expenditure justifies the clinical gain because no maximum cost has been defined. Although models can be used to inform decisions, they must be interpreted and applied carefully. Specifically, the assumption that clinical benefit will persist after biologics are discontinued needs to be validated in order to substantiate the long-term economic value of biologics. More research is also needed to determine the relative economic value of the various biologic agents for specific therapeutic indications.

ACKNOWLEDGMENTS
The authors would like to thank Robert A. Charles, PharmD, MS, for abstracting the data from the literature.

DISCLOSURES
Funding for this study was provided by Amgen, Inc. and was obtained by author Robert W. Dubois. The authors disclose that funding for this research was restricted and that the sponsor was involved with reviewing, editing, and approving the manuscript. Author Chiun-Fang Chiou is employed by Amgen, Inc., and is an Amgen stockholder; he also discloses that he is an author of one of the studies in this subject review. Authors Quan V. Doan and Dubois disclose no potential bias or conflicts of interest relating to this article.
Doan served as principal author of the study. Study concept and design were contributed by all authors. Data collection was the work of Doan; data interpretation was the work of all authors. Writing of the manuscript and its revision were primarily the work of Doan, with input from Dubois and Chiou.