Abstract

Objective. To perform a modelling study on the cost-effectiveness of three outcome-directed strategies in early RA patients: Strategy 1: starting MTX monotherapy, followed by the addition of LEF, followed by MTX with addition of anti-TNF; Strategy 2: start with MTX and LEF combination followed by MTX with anti-TNF; and Strategy 3: immediate start with MTX and anti-TNF.

Methods. A validated Markov model was used to evaluate the cost-effectiveness of the three strategies. Effectiveness of the strategies was determined using daily practice data from two cohorts and used as input parameter in the model. Patients treated according to the strategies were matched for baseline 28-joint DAS (DAS-28). Using Monte Carlo simulation, expected costs, quality-adjusted life-years (QALYs) and incremental cost per QALY gained for a 5-year time horizon were calculated following both a health-care and a societal perspective.

Results. The percentage of patients in remission and number of QALYs were comparable between the three strategies. Starting with a combination (MTX plus LEF or anti-TNF) was more costly than starting with MTX alone. This resulted in an unfavourable incremental cost–effectiveness ratio for starting on anti-TNF vs initially MTX: health-care perspective of €138 028 and from a societal perspective of €136 150 per QALY gained over 5 years.

Conclusion. In this modelling study, starting with MTX or anti-TNF has comparable effectiveness. However, initial anti-TNF was far more expensive than starting with MTX monotherapy. Therefore, based on this study, a treatment strategy starting with MTX monotherapy is favoured over a strategy with MTX and anti-TNF right away in early RA patients.

Introduction

RA is a common idiopathic autoimmune disease characterized by symmetrical synovitis and associated with morbidity and impaired quality of life [1]. The primary goal in early RA is to suppress inflammation as early and completely as possible and ultimately prevent joint damage that leads to pain and functional disability [2–4]. Since RA imposes a substantial economic burden on both the patient and society, it is expected that treatments that decrease functional disability of RA patients would have an effect on the long-term costs of RA as well [5]. Treatment should, therefore, be initiated in the early stages of RA in order to reduce disease activity most effectively: to achieve and sustain clinical remission [1–5]. Moreover, patients should be periodically assessed (3 monthly) for evaluation of disease activity and treatment should be adjusted subsequently [4].

The traditional treatment of early RA involves starting the treatment with DMARDs [6, 7]. Current treatment has shifted towards more intensified treatment—for instance, the use of anti-TNF agents—in order to achieve remission. In early RA, randomized controlled trials (RCTs) showed that anti-TNF agents in combination with MTX are more effective than MTX monotherapy [8, 9]. However, anti-TNF agents are more expensive than DMARDs [10]. Another therapeutic option—the use of combined DMARDs such as MTX and LEF—also resulted in rapidly achieved and sustained low disease activity [11, 12]. Although combinations are more efficacious, such intensive approach is also potentially more toxic and expensive [5]. Moreover, combination therapy might not be necessary for all newly diagnosed RA patients, since 30–40% of the patients will experience a good response to MTX alone [13–15].

In daily practice, a treatment option that is frequently used for early RA patients is to start with MTX, then add a second DMARD if MTX fails and if that fails add an anti-TNF agent. In that case, the relatively expensive anti-TNF treatment is given only to patients having demonstrated failure to DMARDs. However, effective anti-TNF treatment is then delayed. For this reason, it is mandatory to compare both the benefits and the costs of early treatment with anti-TNF agents with strategies that delay use of anti-TNF.

The most robust way to obtain insight into cost and effectiveness of treatment strategies is with an RCT. However, with the exception of the Behandelstrategieën in RA (BeSt) study [16], neither effectiveness nor cost-effectiveness data are available from head-to-head comparisons of these strategies, but there are substantial data available from daily practice registries. Moreover, RCT data do not always reflect clinical practice and observational data could therefore have additional value [17, 18]. Since a modelling study is a good approach to combine several data sources that contain effectiveness and/or costs data [19], modelling was used in order to analyse cost-effectiveness of the proposed strategies [19].

The aim of this study was to compare three different strategies for the treatment of patients with early RA using a cost-effectiveness modelling approach: Strategy 1: starting treatment with MTX monotherapy followed by addition of LEF and then by MTX plus anti-TNF, when patients have inadequate response to treatment; Strategy 2: start with MTX and LEF followed by MTX and anti-TNF; and Strategy 3: start with MTX plus an anti-TNF agent.

Methods

Treatment strategies

The setting of this study is a modelling study that compared cost-effectiveness of three different treatment strategies. In more detail, the three treatment strategies that were compared were: Strategy 1: start with MTX monotherapy, followed by addition of LEF in case of failure, then addition of a first anti-TNF to MTX followed by a second anti-TNF in case of failure and treatment with rituximab in case of failure; Strategy 2: start with the combination of MTX and LEF, followed by MTX and anti-TNF in case of failure and subsequently a second anti-TNF in case of failure and rituximab thereafter; and Strategy 3: start with the combination of MTX and anti-TNF as first therapeutic option, followed by a second anti-TNF and rituximab in case of failure.

MTX was chosen because it is the most commonly used DMARD, alone or in combination [20, 21]. Another treatment approach frequently used is to add SSZ to MTX is case of MTX failure, but it is shown that this approach is not effective [22]. Furthermore, MTX and SSZ share the same pathway, which minimizes the chances of good clinical effect of SSZ when MTX has already failed [23]. LEF has targets in a different pathway [24] and is shown to be effective in MTX failures [11, 12, 25] and was therefore selected as a good alternative for patients who failed MTX. Adalimumab and etanercept were chosen as effective anti-TNF agents since these are the most frequently prescribed anti-TNF agents in the Netherlands. In case of failure on all treatments in the strategy, we have assumed that usual care was provided by a combination of DMARDs. The structure of the model and the treatment strategies are shown in Fig. 1.

Fig. 1

Decision tree for Strategy 1: starting with MTX. Cycle length = 3 months. The decision and chance nodes are represented by ▪ and ○, respectively. The four cut-off points for disease activity are reflected by ◂. High DAS: DAS-28 > 5.1; mod DAS: DAS-28 of >3.2 and ≤5.1; low DAS: DAS-28 of ≤3.2 and >2.6; remission: DAS-28 < 2.6. TNF1: first anti-TNF-α agent (adalimumab or etanercept); TNF2: second anti-TNF-α agent (adalimumab or etanercept); RTX: rituximab; usual: usual care. Strategy 2: initially every patient in MTX + LEF and subsequently similar sequence to Strategy 1. After RTX, patients spent two cycles in MTX and followed by usual care. Strategy 3: initially every patient in TNF1 and subsequently similar sequence to Strategy 1. After RTX, patients spent two cycles in MTX + LEF, two cycles in MTX followed by the usual care.

Fig. 1

Decision tree for Strategy 1: starting with MTX. Cycle length = 3 months. The decision and chance nodes are represented by ▪ and ○, respectively. The four cut-off points for disease activity are reflected by ◂. High DAS: DAS-28 > 5.1; mod DAS: DAS-28 of >3.2 and ≤5.1; low DAS: DAS-28 of ≤3.2 and >2.6; remission: DAS-28 < 2.6. TNF1: first anti-TNF-α agent (adalimumab or etanercept); TNF2: second anti-TNF-α agent (adalimumab or etanercept); RTX: rituximab; usual: usual care. Strategy 2: initially every patient in MTX + LEF and subsequently similar sequence to Strategy 1. After RTX, patients spent two cycles in MTX and followed by usual care. Strategy 3: initially every patient in TNF1 and subsequently similar sequence to Strategy 1. After RTX, patients spent two cycles in MTX + LEF, two cycles in MTX followed by the usual care.

Description of the model

The model builds on a validated Markov model [26]. The patients were initially distributed across several disease states (Markov states) defined by remission, low, moderate and high disease activity according to the 28-joint DAS (DAS-28). It was chosen to define states by DAS-28 because low disease activity defined by DAS-28 is the most important target for treatment, and therefore a DAS-28-based model is a close reflection of daily clinical practice. After the first cycle, patients may be either in remission (initial responders: DAS-28 < 2.6) and remained on their initial treatment for the next 3 months, or not in remission (non-responders: DAS-28 > 2.6) and switched to the next treatment in the strategy. Patients were assumed to sustain remission after being in remission for two cycles, based on study results of early RA patients who were able to sustain their remission (>90%) after remission of two consecutive visits, without therapy adjustment [27].

Effectiveness of treatment was represented by transition probabilities that enable movement to more or less favourable disease states after one cycle [19]. In this model, the cycle length was 3 months and the time horizon was 5 years. The analyses were carried out using TreeAge Pro Software (Williamstown, MA, USA).

Transition probabilities

To calculate transition probabilities, data from effectiveness of the different treatments were needed. Effectiveness data were derived from the following sources: the Nijmegen RA inception cohort and Dutch Rheumatoid Arthritis Monitoring (DREAM) registry [28, 29]. In both cohorts, patients (≥18 years) are included when they have RA according to the 1987 revised ACR criteria for RA [30]. The inception cohort includes patients who have disease duration of <1 year and have no prior DMARD use. The DREAM cohort includes patients who start with their first anti-TNF agent. Patients are regularly assessed at 3-month intervals by an independently trained nurse. Treatment decisions can be made at any time point according to the attending rheumatologist. The local medical ethics committee (Commissie Mensgebonden Onderzoek, Region Arnhem—Nijmegen, The Netherlands) approved the study. All patients gave their informed consent before inclusion in the inception cohort.

From these data sources, we selected all patients who received one of the strategies in the period 2003–08 from the two cohorts. After inclusion, patients with complete assessments of disease activity at baseline, 3 and 6 months were used to determine effectiveness. In case of missing ESR values, the mean of the previous and following values was used (linear intrapolation) for imputation and to calculate the DAS-28. The occurrence of DAS-28 Markov states after treatment was used to calculate transition probabilities (occurrence of a DAS-28 Markov state divided by the total treatment group) (see supplementary data available at Rheumatology Online). Since we used data from daily practice, some treatment groups were actually low, i.e. MTX plus LEF.

Data about MTX and/or LEF treatment were obtained from early RA patients as included in the Nijmegen inception cohort and data about treatment of anti-TNF agents and rituximab were derived from the DREAM cohort. Efficacy of MTX was assessed in MTX-naïve patients who had a dosage of at least 15 mg/week. For efficacy of LEF, anti-TNF and rituximab standard dosage, schedules were used as given in the Dutch medication registration (see supplementary data available at Rheumatology Online) [31]. Assuming that response to a second anti-TNF agent is less effective than the response to a first anti-TNF agent [13, 32], efficacy data were based on using a first and a second anti-TNF agent, respectively. According to the daily practice, patients received at least two different DMARDs before anti-TNF agents. For the usual care data, patients received a combination of DMARDs that encompassed MTX and another DMARD. In all available datasets, patients were treated with CSs at the discretion of the treating physicians; the percentage of patients on CSs was ∼30% and the same in the used datasets. It was therefore not necessary to control for the use of CS in the analyses.

The selection of patients treated according to the different strategies might have led to differences in baseline characteristics between the three groups of patients. If these baseline characteristics are prognostic for the effectiveness of therapy, it can cause confounding. The DAS-28 at baseline is known to be a strong and consistent predictor of treatment response [33]. Therefore, patients that started with MTX, MTX plus LEF or anti-TNF agents were matched using their DAS-28 at baseline. This was done by category matching in which patients available in the datasets were categorized into 1 of 10 categories according to their baseline DAS-28. Random samples of patients were drawn from each category using a ratio based on the initial sample sizes in the available datasets of one MTX patient : one LEF patient : three anti-TNF patients. Differences in baseline characteristics (after matching) were tested by using one-way analyses of variance and chi-squared tests where appropriate. The transition probabilities as calculated in the described daily clinical practice datasets were then entered in the Markov model. This led to the distribution of hypothetical patients over the four states of disease activity as defined in the model.

Cost and utilities

Utility and costs were related to the four states of disease activity in the model of this study. The relationship between these states and utility and costs had been previously validated in the Markov model of Welsing et al. [26]. All utility values and cost that were related to the Markov states were measured in a previous 48-week multicentre trial of MTX treatment in RA patients [34]. In this trial, costs were measured using a questionnaire and a patient diary and utility values were collected from the European Quality of Life-5 dimensions (EuroQoL-5D) questionnaire, valued using the British tariff. In addition, the mean costs and utility were calculated for patients in each Markov state, as defined by their mean disease activity over the past 3 months (costs) or at Week 48 (utility).

For the calculations of the costs, two different perspectives were applied. First, the health-care perspective, including medical costs: consultations of general practitioner, outpatient visits to the rheumatologist and other medical specialists, surgery, hospitalization and other co-medication. Second, the societal perspective, including non-medical costs: absence from paid labour valued according to the friction cost method, travel expenses and out-of-pocket costs (alternative treatment, home help and expenses related to RA) [34].

Cost prices for each drug treatment were added separately and derived from the Dutch National tariff list using standard dosage schedules according to the Dutch registration (see supplementary data available at Rheumatology Online) [31]. In case of sustained treatment with rituximab (>6 months), a mean cost of €1817/cycle was used, assuming a mean time of 9 months between treatments [35].

Cost–effectiveness analysis

Cost–effectiveness analysis (CEA) was performed using second-order Monte Carlo simulation incorporating the influence of parameter uncertainty on the expected outcomes: probabilistic sensitivity analysis [19]. During simulation, individual hypothetical patients were sampled based on the distributions of transition probabilities. This was simulated for 1000 patients. Uncertainties in all input parameters used in the Markov model, including costs, utility and transition probabilities, were considered by means of distributions [36]. Dirichlet distributions (groups of beta distributions for integer forms) represented transition probabilities. These distributions were mainly determined by the total size of a treatment group; a small group relates to more uncertainty. Beta distributions (real number form) and gamma distributions were used for utility values and costs, respectively. Applying Monte Carlo simulation, a value from the distribution (with the chance depending on the distribution) was drawn from all input parameters during each one of the 1000 simulations. Therefore, the expected outcomes of the model were also represented by distributions, representing the uncertainty of the outcomes based on the parameter uncertainty.

Outcome measures

The model was evaluated from a health-care and a societal perspective [19]. Effects and costs were discounted at 4%/year resulting in 1%/cycle [19]. The primary outcome measure was the incremental cost–effectiveness ratio (ICER) [37]. The ICER was always compared with the first strategy as a reference strategy. Acceptability curves were derived illustrating, in a Bayesian fashion with an uninformative prior, the probability of being cost-effective given a certain threshold [willingness to pay (WTP)] for a quality-adjusted life-year (QALY) [19]. Percentages of remission at the end of simulation were reported as a secondary outcome.

Scenario analysis

The base-case analysis calculated cost and effectiveness using the transition probabilities with remission (DAS-28 <2.6) as outcome of response. In order to investigate the impact of a less strict outcome of response, we performed a scenario analysis concerning low DAS-28 (≤3.2) as cut-off point to hold the same treatment or to move to another. For this purpose, transition probabilities were again calculated from the Nijmegen inception and DREAM cohorts defined as the probability to achieve low disease activity, and accordingly, Monte Carlo simulation was repeated.

For this model, we used data from patients who previously used two DMARDs to model the effectiveness of anti-TNF agents. We assumed that this effectiveness was the same for patients who started the strategy with anti-TNF agents. However, this could have led to an underestimation of the real effectiveness since clinical trials using early RA patients showed higher treatment responses than trials with established RA patients [3, 8, 61]. To get insight into the effect of this underestimation, we performed a second scenario analysis. In this analysis, remission results of anti-TNF treatment as shown in the COMET trial in DMARD-naïve patients with early RA were used [3]. This trial showed that 30% of the early RA patients achieved remission after 3 months of treatment with anti-TNF.

Results

Baseline characteristics

After matching for disease activity, there were no significant differences in baseline DAS-28 between the treatment strategies (Table 1). On the other hand, a difference in disease duration remained between the treatment strategies. As expected from treatment in daily practice, patients received two or more DMARDs before their treatment with anti-TNF agents.

Table 1

Demographic and baseline disease characteristics of the patient populations for the input dataa

 MTX (n = 112) MTX + LEF (n = 47) Anti-TNF (n = 332) P-value 
Age, mean (s.d.), years 64 (14) 60 (14) 57 (13) 0.388 
Women, % 70 62 70 0.231 
RF positive, % 77 82 76 0.776 
Disease duration, median (IQR), years 1.0 (0.4–4.6) 2.0 (1.2–8.6) 7.1 (2.7–13.2) 0.000 
DAS-28, mean (s.d.4.9 (1.2) 4.7 (1.1) 5.1 (1.3) 0.309 
Dosage,b median (IQR) 15 (15–20) 20 40/50 – 
Previous DMARDs, % 58 62 100 0.019 
No. of DMARDs used, median (IQR) 1 (1–2) 2 (1–2) 2 (2–3) 0.000 
 MTX (n = 112) MTX + LEF (n = 47) Anti-TNF (n = 332) P-value 
Age, mean (s.d.), years 64 (14) 60 (14) 57 (13) 0.388 
Women, % 70 62 70 0.231 
RF positive, % 77 82 76 0.776 
Disease duration, median (IQR), years 1.0 (0.4–4.6) 2.0 (1.2–8.6) 7.1 (2.7–13.2) 0.000 
DAS-28, mean (s.d.4.9 (1.2) 4.7 (1.1) 5.1 (1.3) 0.309 
Dosage,b median (IQR) 15 (15–20) 20 40/50 – 
Previous DMARDs, % 58 62 100 0.019 
No. of DMARDs used, median (IQR) 1 (1–2) 2 (1–2) 2 (2–3) 0.000 

aData are obtained from the Nijmegen RA inception cohort and the DREAM registry. bDosage MTX mg/week, LEF mg/day and anti-TNF was split into: adalimumab with a dosage schedule of 40 mg every other week and etanercept with a dosage schedule of 50 mg every week.

CEA

ICER

The estimated mean costs and effectiveness of each treatment strategy after 5 years according to the health-care or societal perspective are shown in Table 2. This table shows that the number of QALYs in all strategies was comparable; however, starting with MTX and anti-TNF (€17 574) was more costly than MTX alone (€16 620). The additional cost for the anti-TNF strategy compared with the MTX strategy resulted in an ICER per patient of €138 056 (95% CI €137 007, €139 123) according to the health-care perspective and €136 207 (95% CI €135 022, €137 363) according to the societal perspective per QALY gained over 5 years. The uncertainty in the decision was small as was shown by the steep acceptability curve (data not shown), in accordance with the relatively small 95% CIs of the ICERs.

Table 2

Costs, number of QALYs, CERs and ICERs per patient over 5 years for each treatment strategy using remission and low disease activity as response ratea

Outcome MTX monotherapy MTX + LEF combination MTX + anti-TNF 
Base-case analyses: remission    
    Health-care perspectiveb    
    Total costs, € 16 620 (16 607–16 633) 18 313 (18 301–18 327) 17 574 (17 574–17 588) 
    Total no. of QALYs 3.086 (3.079–3.092) 3.089 (3.083–3.096) 3.093 (3.086–3.099) 
    ICER, €/QALYc  438 056 (434 536–441 697) 138 056 (137 007–139 123) 
 Societal perspectiveb    
    Total costs, € 17 580 (17 558–17 601) 19 269 (19 247–19 290) 18 521 (18 499–18 542) 
    Total no. of QALYs 3.086 (3.079–3.092) 3.089 (3.083–3.096) 3.093 (3.086–3.099) 
    ICER, €/QALY  437 014 (433 137–441 153) 136 207 (135 022–137 363) 
Scenario analyses: low disease activity    
 Health-care perspectiveb    
    Total costs, € 21 780 (21 766–21 792) 26 022 (26 009–26 035) 28 346 (28 333–28 359) 
    Total no. of QALYs 3.187 (3.179–3.196) 3.193 (3.185–3.202) 3.200 (3.192–3.209) 
    ICER €/QALYc  742 508 (736 295–748 616) 512 355 (508 210–516 401) 
 Societal perspectiveb    
    Total costs, € 23 000 (22 964–23 041) 27 243 (27 205–27 284) 29 560 (29 523–29 602) 
    Total no. of QALYs 3.187 (3.179–3.196) 3.193 (3.184–3.202) 3.200 (3.192–3.209) 
    ICER, €/QALY  742 981 (735 942–749 816) 512 180 (507 565–516 682) 
Scenario analyses: more effectiveness of anti-TNF    
 Health-care perspectiveb    
    Total costs, € 16 618 (16 606–16 632) 18 312 (18 300–18 326) 20 507 (20 496–20 520) 
    Total no. of QALYs 3.086 (3.080–3.092) 3.089 (3.084–3.096) 3.119 (3.113–3.126) 
    ICER, €/QALYc  437 785 (434 272–441 619) 116 598 (115 556–117 619) 
 Societal perspectiveb    
    Total costs, € 17 580 (17 558–17 603) 19 269 (19 246–19 292) 21 412 (21 389–21 433) 
    Total no. of QALYs 3.086 (3.080–3.092) 3.089 (3.084–3.096) 3.119 (3.113–3.126) 
    ICER, €/QALY  436 965 (432 724–441 049) 114 982 (113 768–116 076) 
Outcome MTX monotherapy MTX + LEF combination MTX + anti-TNF 
Base-case analyses: remission    
    Health-care perspectiveb    
    Total costs, € 16 620 (16 607–16 633) 18 313 (18 301–18 327) 17 574 (17 574–17 588) 
    Total no. of QALYs 3.086 (3.079–3.092) 3.089 (3.083–3.096) 3.093 (3.086–3.099) 
    ICER, €/QALYc  438 056 (434 536–441 697) 138 056 (137 007–139 123) 
 Societal perspectiveb    
    Total costs, € 17 580 (17 558–17 601) 19 269 (19 247–19 290) 18 521 (18 499–18 542) 
    Total no. of QALYs 3.086 (3.079–3.092) 3.089 (3.083–3.096) 3.093 (3.086–3.099) 
    ICER, €/QALY  437 014 (433 137–441 153) 136 207 (135 022–137 363) 
Scenario analyses: low disease activity    
 Health-care perspectiveb    
    Total costs, € 21 780 (21 766–21 792) 26 022 (26 009–26 035) 28 346 (28 333–28 359) 
    Total no. of QALYs 3.187 (3.179–3.196) 3.193 (3.185–3.202) 3.200 (3.192–3.209) 
    ICER €/QALYc  742 508 (736 295–748 616) 512 355 (508 210–516 401) 
 Societal perspectiveb    
    Total costs, € 23 000 (22 964–23 041) 27 243 (27 205–27 284) 29 560 (29 523–29 602) 
    Total no. of QALYs 3.187 (3.179–3.196) 3.193 (3.184–3.202) 3.200 (3.192–3.209) 
    ICER, €/QALY  742 981 (735 942–749 816) 512 180 (507 565–516 682) 
Scenario analyses: more effectiveness of anti-TNF    
 Health-care perspectiveb    
    Total costs, € 16 618 (16 606–16 632) 18 312 (18 300–18 326) 20 507 (20 496–20 520) 
    Total no. of QALYs 3.086 (3.080–3.092) 3.089 (3.084–3.096) 3.119 (3.113–3.126) 
    ICER, €/QALYc  437 785 (434 272–441 619) 116 598 (115 556–117 619) 
 Societal perspectiveb    
    Total costs, € 17 580 (17 558–17 603) 19 269 (19 246–19 292) 21 412 (21 389–21 433) 
    Total no. of QALYs 3.086 (3.080–3.092) 3.089 (3.084–3.096) 3.119 (3.113–3.126) 
    ICER, €/QALY  436 965 (432 724–441 049) 114 982 (113 768–116 076) 

Data are presented as median (2.5–97.5 percentile) of the cost (€) and QALYs from all 1000 simulations in the probabilistic sensitivity analysis. Costs and utilities were discounted 4% per year. aRemission: DAS-28 < 2.6; low disease activity: DAS-28 ≤ 3.2. bAccording to the health-care perspective, medical costs consisted of physician consultations, surgery, etc. and drug treatment. According to the societal perspective, total costs were composed of medical and non-medical (absence from paid labour, travel expenses, etc.) costs. cAn ICER is the additional costs of a treatment as compared with an alternative (MTX monotherapy) divided by the additional effect of a treatment compared with that of the alternative treatment (MTX monotherapy).

Also, the strategy that started with MTX and LEF was more costly than the strategy that started with MTX alone, without any extra effect. Since the MTX–LEF strategy costs more than the anti-TNF strategy and results in less effectiveness, the MTX–LEF strategy was excluded from ICER calculation based on extended dominance.

First scenario analysis

The first scenario analysis showed that changing the response rate to low disease activity resulted in an ICER of €512 355 (from the health-care perspective) and €512 180 (from the societal perspective) per QALY per patient over 5 years for Strategy 3 vs Strategy 1 (Table 2). The strategy of starting with MTX–LEF combination remained dominated in this analysis since the incremental cost for Strategy 2 vs Strategy 1 was €742 271 (from the health-care perspective) and €742 675 (from the societal perspective) per QALY per patient gained over 5 years.

Second scenario analysis

In the second scenario analysis, an estimate of 30% of the DMARD-naïve patients achieving remission with anti-TNF [3] was used as input for the model in comparison with the estimate of 20% of patients who used two DMARDs before anti-TNF treatment as was used as input for the base-case analysis. The effectiveness in terms of the expected number of QALYs per patient over 5 years was 3.08 in the MTX strategy and 3.12 in the anti-TNF strategy (Table 2). After 5 years, costs (health-care perspective) were €16 619 and €20 508 for the MTX strategy and anti-TNF strategy, respectively. According to the societal perspective, these were €17 581 and €21 412 for the MTX strategy and anti-TNF strategy, respectively. This scenario analysis resulted in a more favourable ICER of the anti-TNF agents treatment strategy: €116 598 (95% CI €115 556, €117 619) from the health-care perspective and €114 982 (95% CI €113 768, €116 076) from the societal perspective (Table 2). However, these ICERs are still considered high (more than the generally accepted threshold value of €100 000 [38]) and starting with MTX remained, therefore, the most cost-effective option over Strategy 3. Furthermore, the strategy that started with MTX and LEF was more costly and lay far above €100 000 with a WTP of €437 960 per QALY.

Remission

The percentages of patients in the remission states after treatment of all strategies are shown in Table 3. Of the patients who started with the combination of MTX and anti-TNF, a relatively high number attained remission after this first treatment (15%). In the MTX strategy, a lower percentage of patients achieved remission (7%) after their first treatment of MTX. After 10 cycles (2.5 years), in each treatment strategy equilibrium was reached, meaning that there were no more transitions between Markov states. Moreover, there was no difference in the total number of patients who were in remission between the three strategies; 38% of patients had sustained remission. The latter finding is also in agreement with the fact that effectiveness in terms of QALYs gained was similar between the three strategies (as already shown in Table 2).

Table 3

Percentage of patients in remission states (DAS-28 < 2.6) for each treatment strategy as a result of the 5-year base-case Markov simulation

 MTX Sequencea MTX + LEF Sequencea MTX + anti-TNF Sequencea 
MTX monotherapy 6.52 4.32 4.32 
MTX + LEF 7.23 7.74 5.55 
First anti-TNF 13.34 14.27 15.46 
Second anti-TNF 5.32 5.69 6.16 
Rituximab 5.76 6.16 6.68 
 MTX Sequencea MTX + LEF Sequencea MTX + anti-TNF Sequencea 
MTX monotherapy 6.52 4.32 4.32 
MTX + LEF 7.23 7.74 5.55 
First anti-TNF 13.34 14.27 15.46 
Second anti-TNF 5.32 5.69 6.16 
Rituximab 5.76 6.16 6.68 

aSequence number of DMARDs or anti-TNF given within each treatment strategy.

Costs

Table 2 shows that there is no relevant difference in efficacy in terms of QALYs gained between all three strategies. Therefore, the unfavourable ICERs obtained for Strategy 3 vs Strategy 1 can largely be attributed to the cost of both the strategies. This is also demonstrated in Fig. 2A, which depicts the medical cost histograms of the three strategies, according to the health-care perspective. It is shown that there is a large overlap in medical cost between the anti-TNF strategy and the MTX strategy. The difference in expected medical cost between these two strategies after 5 years is actually very small (€954). Using the societal perspective instead of the health-care perspective, the MTX strategy was still the least expensive strategy despite the overlap of the anti-TNF strategy (data not shown).

Fig. 2

Histograms according to a health-care perspective of starting with MTX monotherapy (Strategy 1) vs starting with MTX and LEF therapy (Strategy 2) and starting with an anti-TNF agent therapy (adalimumab or etanercept) (Strategy 3) using remission as response criteria (A) or low disease activity as response criteria (B). Histograms are generated from 1000 simulations in the probabilistic sensitivity analysis with a time-horizon of 5 years. The horizontal axis depicts total medical costs per QALY after 5 years. The vertical axis depicts number of patients simulated.

Fig. 2

Histograms according to a health-care perspective of starting with MTX monotherapy (Strategy 1) vs starting with MTX and LEF therapy (Strategy 2) and starting with an anti-TNF agent therapy (adalimumab or etanercept) (Strategy 3) using remission as response criteria (A) or low disease activity as response criteria (B). Histograms are generated from 1000 simulations in the probabilistic sensitivity analysis with a time-horizon of 5 years. The horizontal axis depicts total medical costs per QALY after 5 years. The vertical axis depicts number of patients simulated.

The analysis including low disease activity as clinical outcome, gave similar results compared with that of using remission as is shown in Fig. 2B: the MTX strategy remained the least expensive option. However, starting with anti-TNF agents was far more costly than starting with MTX monotherapy and even than MTX plus LEF. This resulted in a large difference in medical cost, according to the health-care perspective, between Strategies 1 and 3 over 5 years (€6566).

Discussion

According to the results of this study, starting with the combination of MTX and anti-TNF agents was less cost-effective than starting with MTX monotherapy in early RA. This was explained by the relatively comparable effectiveness between the two strategies and lower costs when starting with MTX monotherapy. Additionally, in the model, reaching remission (DAS-28 < 2.6) was used as response criterion, a rather strict criterion. Accordingly, most patients starting with MTX monotherapy failed this therapy and were already on anti-TNF after a few cycles. Moreover, all three strategies reached their equilibrium after 10 cycles. On the other hand, a number of patients were successfully treated with MTX alone and did not need to be treated with anti-TNF, resulting in lower costs of the MTX-starting strategy. Taken together, the small difference in QALYs gained compared with higher costs, resulted in a large ICER for starting with the combination of MTX and anti-TNF vs starting with MTX alone: €138 028 according to the health-care perspective and €136 150 according to the societal perspective. These ICERs indicate that €138 028 or €136 150 is needed to invest in a gain of one QALY over 5 years with the anti-TNF strategy. These ICERs are generally considered as unfavourable ICERs as only ICERs below the threshold value of €100 000 are considered potentially cost-effective [38].

In the case of changing the response criterion to low disease activity (DAS-28 ≤ 3.2), a less strict criterion (first scenario analysis), more patients were effectively treated with MTX monotherapy against lower costs of this strategy. Subsequently, initial MTX became an even more preferable option over initial MTX and anti-TNF. In the base-case analysis, efficacy data of anti-TNF were derived from patients with prior DMARD use. This may have led to an underestimation of the real efficacy of anti-TNF in DMARD-naïve patients. Therefore, a second scenario analysis was applied assuming higher effectiveness of anti-TNF in early RA patients based on the COMET trial [3, 8]. This analysis led to greater effectiveness of starting with anti-TNF, and thus a higher number of QALYs gained was obtained. This resulted in a more favourable ICER of MTX and anti-TNF against MTX. However, this ICER was still considered unfavourable (€116 598).

When taking costs into consideration and based on this CEA, there seems to be no extra benefit of starting with the combination of MTX and anti-TNF agents; all ICERs obtained were higher than the threshold value of €100 000. These results are in accordance with results of the CEA from the BeSt study. This study concluded that initial treatment with infliximab and MTX gained significantly more QALYs, but was more costly compared with initial MTX and was, therefore, considered not to be cost-effective using a societal cost perspective according to the friction cost method. However, if productivity was valued according to the human capital method, then starting with infliximab was more cost-effective than starting with DMARDs [16]. But the friction cost method allows a more realistic estimate of productivity costs by incorporating the possibility of replacement of absenteeism, which strengthens our choice for this method [39]. We were not able to repeat our CEA according to the human capital method as cost data were used from a previous publication in which they were already valued according to the friction cost method. Altogether, it can be concluded from the CEA of the BeSt study that both an MTX step-up strategy or MTX combination with high-dose prednisone [the Combinatietherapie Bij Rheumatoide Artritis (COBRA) scheme] are preferred over initial treatment with infliximab in newly diagnosed RA patients. Furthermore, a systematic review of CEAs of anti-TNF agents concluded that anti-TNF agents are most cost-effective when used as last therapy, i.e. after failure of DMARDs [40]. Our data also confirm a recent cost-effectiveness modelling study by Finckh et al. [41] which concludes that the additional costs of very early intervention with biologics may not be justified and that strategies involving early conventional DMARDs are preferred.

The real challenge for the future is how we can influence the ICER for the use of anti-TNF compared with DMARD strategies? Two important issues are involved with this challenge. First, the costs of drugs, key drivers of medical costs in the case of anti-TNF therapy, might be changed when drug-free remission becomes a realistic option. There is evidence to suggest that drug-free remission especially in recent-onset RA is realistic as well as anti-TNF-free remission after initial early use of anti-TNF [27, 42]. Secondly, the cost–effectiveness ratio (CER) might be influenced by taking other outcomes into account, such as joint damage or functional capacity, which determine labour participation and surgery in particular. Additionally, there is evidence that in patients with early RA, anti-TNF results in less radiographic progression than MTX monotherapy, which may decrease long-term cost as well and may lead to more favourable CERs of anti-TNF [3, 15, 43]. However, long-term outcomes of the positive effects of anti-TNF in terms of preventing joint damage has not yet been established [44]. Therefore, in this study we focused on clinical efficacy (DAS-28) only and consequently, long-term outcome in terms of joint damage remains a matter of further research. From another point of view, this study showed low remission rates after MTX, which might actually become higher in the case of optimal dosage schemes. Accordingly, ICERs that are even more favourable could be obtained with a strategy of starting with MTX alone.

Though this study examined the (common) strategy of starting with MTX and/or LEF, there remain, however, additional strategies for further research, i.e. starting with triple therapy would be very interesting to assess as well. In addition, the recent Treatment of Early Aggressive Rheumatoid Arthritis (TEAR) study [45] showed us that initial triple therapy, a less costly DMARD combination of MTX, SSZ and plaquenil, is more effective than starting MTX alone and may (hypothetically) result in an even more favourable ICER of Strategy 1 (DMARDs started) vs Strategy 3 (anti-TNF started). The COBRA strategy [combination of MTX, SSZ and high-dose prednisone (stepped down)] has shown long-term effects on progression of joint damage and mortality, and therefore could be a good alternative for initial TNF treatment [46].

Recently, Markov models have been used extensively in RA. The structures of these models differ on the measure used to define the Markov states. A frequently used measure is the HAQ [47–51] as opposed to the DAS-28 as used in our model. The HAQ as well as the DAS-28 comprises an absolute measure of clinical importance in RA, but achieving low DAS-28 is the most important target for treatment in RA. Furthermore, the HAQ is defined by psychosocial factors [52, 53] and the HAQ is largely influenced by age and the presence of comorbidities [54]. Therefore, the relation of the HAQ to costs and utility might not be the same in different phases of the disease and in different populations. A clinically relevant division in the HAQ to define Markov states is less clear, which is also apparent from the different definitions within the Markov models using the HAQ. For the DAS, clear cut-off values for remission, low disease activity, moderate disease activity and high disease activity are present [55].

A reference case for economic evaluations in RA was proposed by the Economic Working Group of OMERACT [56]. Our study is in accordance with all recommendations except the one considering mortality as an outcome measure, which was not accounted for in this study. However, a difference in mortality between the treatment strategies is not to be expected within the time horizon of 5 years as in our analyses.

A modelling study has its limitations. The model in this study used efficacy data from the Nijmegen inception cohort and DREAM registry, both large registries of RA patients in the Netherlands. Cohort data have the advantage of being closely related to daily practice care and, therefore, the patients modelled in this study are supposed to be representative of the general RA population attending outpatient clinics. However, by using observational data, the strategies in this CEA were indirectly compared, with possible differences in baseline characteristics between the groups. In order to have comparable groups at baseline, we matched for disease activity at baseline, the most important prognostic factor for clinical efficacy [33], and showed that these were comparable after matching. Though disease duration was different between the strategies, this would not have changed the results since disease duration is not a prognostic factor for clinical efficacy according to a recent systematic review of 18 studies on predictors for achieving remission in RA [33]. Besides baseline characteristics, differences in CS use may contribute to different treatment outcomes. However, the percentage of patients on CSs was comparable in all treatment strategies (∼30%). Therefore, it was not necessary to control for the use of CSs.

We note that absolute comparability between different treatments is only expected through randomization [57]. However, no more than the BeSt study performed a direct comparison of four treatment strategies [15]. The ideal CEA would use data from a large, long-term RCT without selecting patients, examining both efficacy and costs of different treatment strategies in early RA patients. Such head-to-head comparisons of treatment strategies are scarce, being costly and sometimes not feasible. Consequently, a modelling study has an important contribution to select relevant treatment strategies and can be seen as a pilot of the ideal head-to-head study.

Since we used observational data, the number of patients with specific treatments available in the data set (DASs) was sometimes low. This can result in large uncertainty in the input parameter. We performed probability sensitivity analysis that considered uncertainty by using distributions of the efficacy data and took the number of patients into account (transition probabilities). Further, we have shown that the ICERs given were actually accompanied by small 95% CIs indicating a low level of uncertainty.

Several assumptions were applied in the Markov model. First, a cycle length of 3 months was assumed and accordingly treatment responders (remission) were identified after one cycle. It could be argued that a 3-monthly cycle length may be too short and remission as treatment outcome may be very strict. However, 3-monthly assessments are in accordance with clinical practice and aiming at remission is currently regarded as a treatment goal [4, 58, 59]. Secondly, we assumed that clinical responses of anti-TNF agents in DMARD-naïve RA patients were comparable to those who failed DMARDs. This assumption was applied for practical reasons: efficacy data of anti-TNF were only available from patients who failed at least two DMARDs. This may have led to lower treatment responses since clinical trials with early RA patients showed higher responses than trials with established RA patients [8, 60, 61]. Therefore, we carried out a second scenario analysis in case of a greater effectiveness of anti-TNF in DMARD-naïve patients [3].

Keeping these limitations in mind, we conclude that starting with MTX or anti-TNF showed comparable effectiveness in terms of disease activity and QALYs. However, initial anti-TNF was far more expensive than starting with MTX monotherapy. Therefore, based on this study a treatment strategy starting with MTX monotherapy is favoured over a strategy with MTX and anti-TNF or MTX and LEF right away in early RA.

graphic

Supplementary data

Supplementary data are available at Rheumatology Online.

Acknowledgements

The work of the first author (L.G.S.) is supported by a grant from Wyeth Pharmaceuticals. This funding body did not have any contribution to study design; data collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication.

Funding: Funding of the Dutch National Health Insurance Board and the Dutch affiliations of Wyeth Pharmaceuticals, Abbott Pharmaceuticals, Schering-Plough Corporation, Roche Pharmaceuticals and Bristol-Myers Squibb enabled the data collection for the DREAM cohort.

Disclosure statement: P.L.C.M.vR. has received honoraria for lecturing and research grants from MSD, Wyeth, Abbott, Roche and BMS. All other authors have declared no conflicts of interest.

References

1
Breedveld
FC
Current and future management approaches for rheumatoid arthritis
Arthritis Res
 , 
2002
, vol. 
4
 
Suppl. 2
(pg. 
S16
-
21
)
2
Smolen
JS
Aletaha
D
Machold
KP
Therapeutic strategies in early rheumatoid arthritis
Best Pract Res Clin Rheumatol
 , 
2005
, vol. 
19
 (pg. 
163
-
77
)
3
Emery
P
Breedveld
FC
Hall
S
, et al.  . 
Comparison of methotrexate monotherapy with a combination of methotrexate and etanercept in active, early, moderate to severe rheumatoid arthritis (COMET): a randomised, double-blind, parallel treatment trial
Lancet
 , 
2008
, vol. 
372
 (pg. 
375
-
82
)
4
van Tuyl
LH
Lems
WF
Voskuyl
AE
, et al.  . 
Tight control and intensified COBRA combination therapy in early rheumatoid arthritis: 90% remission in a pilot trial
Ann Rheum Dis
 , 
2008
, vol. 
67
 (pg. 
1574
-
7
)
5
Yelin
E
Wanke
LA
An assessment of the annual and long-term direct costs of rheumatoid arthritis: the impact of poor function and functional decline
Arthritis Rheum
 , 
1999
, vol. 
42
 (pg. 
1209
-
18
)
6
Combe
B
Early rheumatoid arthritis: strategies for prevention and management
Best Pract Res Clin Rheumatol
 , 
2007
, vol. 
21
 (pg. 
27
-
42
)
7
Choy
EH
Smith
C
Dore
CJ
Scott
DL
A meta-analysis of the efficacy and toxicity of combining disease-modifying anti-rheumatic drugs in rheumatoid arthritis based on patient withdrawal
Rheumatology
 , 
2005
, vol. 
44
 (pg. 
1414
-
21
)
8
Breedveld
FC
Weisman
MH
Kavanaugh
AF
, et al.  . 
The PREMIER study: a multicenter, randomized, double-blind 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
, vol. 
54
 (pg. 
26
-
37
)
9
Bathon
JM
Martin
RW
Fleischmann
RM
, et al.  . 
A comparison of etanercept and methotrexate in patients with early rheumatoid arthritis
N Engl J Med
 , 
2000
, vol. 
343
 (pg. 
1586
-
93
)
10
Michaud
K
Messer
J
Choi
HK
Wolfe
F
Direct medical costs and their predictors in patients with rheumatoid arthritis: a three-year study of 7,527 patients
Arthritis Rheum
 , 
2003
, vol. 
48
 (pg. 
2750
-
62
)
11
Kremer
JM
Genovese
MC
Cannon
GW
, et al.  . 
Concomitant leflunomide therapy in patients with active rheumatoid arthritis despite stable doses of methotrexate. A randomized, double-blind, placebo-controlled trial
Ann Intern Med
 , 
2002
, vol. 
137
 (pg. 
726
-
33
)
12
Strand
V
Cohen
S
Schiff
M
, et al.  . 
Treatment of active rheumatoid arthritis with leflunomide compared with placebo and methotrexate. Leflunomide Rheumatoid Arthritis Investigators Group
Arch Intern Med
 , 
1999
, vol. 
159
 (pg. 
2542
-
50
)
13
Klareskog
L
van der Heijde
D
de Jager
JP
, et al.  . 
Therapeutic effect of the combination of etanercept and methotrexate compared with each treatment alone in patients with rheumatoid arthritis: double-blind randomised controlled trial
Lancet
 , 
2004
, vol. 
363
 (pg. 
675
-
81
)
14
van der Heijde
D
Klareskog
L
Rodriguez-Valverde
V
, et al.  . 
Comparison of etanercept and methotrexate, alone and combined, in the treatment of rheumatoid arthritis: two-year clinical and radiographic results from the TEMPO study, a double-blind, randomized trial
Arthritis Rheum
 , 
2006
, vol. 
54
 (pg. 
1063
-
74
)
15
Goekoop-Ruiterman
YP
de Vries-Bouwstra
JK
Allaart
CF
, 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
, vol. 
52
 (pg. 
3381
-
90
)
16
Van den Hout
WB
Goekoop-Ruiterman
YP
Allaart
CF
, et al.  . 
Cost-utility analysis of treatment strategies in patients with recent-onset rheumatoid arthritis
Arthritis Rheum
 , 
2009
, vol. 
61
 (pg. 
291
-
9
)
17
Kievit
W
Fransen
J
Oerlemans
AJ
, et al.  . 
The efficacy of anti-TNF in rheumatoid arthritis, a comparison between randomized controlled trials and clinical practice
Ann Rheum Dis
 , 
2007
, vol. 
66
 (pg. 
1473
-
8
)
18
Zink
A
Strangfeld
A
Schneider
M
, et al.  . 
Effectiveness of tumor necrosis factor inhibitors in rheumatoid arthritis in an observational cohort study: comparison of patients according to their eligibility for major randomized clinical trials
Arthritis Rheum
 , 
2006
, vol. 
54
 (pg. 
3399
-
407
)
19
Hunink
MG
Glasziou
PP
Siegel
JE
, et al.  . 
Decision making in health and medicine: integrating evidence and values
 , 
2001
2nd edition
Cambridge
Cambridge University Press
20
Newsome
G
Guidelines for the management of rheumatoid arthritis: 2002 update
J Am Acad Nurse Pract
 , 
2002
, vol. 
14
 (pg. 
432
-
7
)
21
O’Dell
JR
Therapeutic strategies for rheumatoid arthritis
N Engl J Med
 , 
2004
, vol. 
350
 (pg. 
2591
-
602
)
22
Haagsma
CJ
van Riel
PL
de Jong
AJ
van de Putte
LB
Combination of sulphasalazine and methotrexate versus the single components in early rheumatoid arthritis: a randomized, controlled, double-blind, 52 week clinical trial
Br J Rheumatol
 , 
1997
, vol. 
36
 (pg. 
1082
-
8
)
23
Haagsma
CJ
Russel
FG
Vree
TB
, et al.  . 
Combination of methotrexate and sulphasalazine in patients with rheumatoid arthritis: pharmacokinetic analysis and relationship to clinical response
Br J Clin Pharmacol
 , 
1996
, vol. 
42
 (pg. 
195
-
200
)
24
Breedveld
FC
Dayer
JM
Leflunomide: mode of action in the treatment of rheumatoid arthritis
Ann Rheum Dis
 , 
2000
, vol. 
59
 (pg. 
841
-
9
)
25
Osiri
M
Shea
B
Robinson
V
, et al.  . 
Leflunomide for treating rheumatoid arthritis
Cochrane Database Syst Rev
 , 
2003
, vol. 
1
 pg. 
CD002047
 
26
Welsing
PM
Severens
JL
Hartman
M
, et al.  . 
The initial validation of a Markov model for the economic evaluation of (new) treatments for rheumatoid arthritis
Pharmacoeconomics
 , 
2006
, vol. 
24
 (pg. 
1011
-
20
)
27
van der Kooij
SM
Goekoop-Ruiterman
YP
de Vries-Bouwstra
JK
, et al.  . 
Probability of continued low disease activity in patients with recent onset rheumatoid arthritis treated according to the disease activity score
Ann Rheum Dis
 , 
2008
, vol. 
67
 (pg. 
266
-
9
)
28
Welsing
PM
van Riel
PL
The Nijmegen inception cohort of early rheumatoid arthritis
J Rheumatol
 , 
2004
, vol. 
69
 
Suppl.1
(pg. 
14
-
21
)
29
Kievit
W
Adang
EM
Fransen
J
, et al.  . 
The effectiveness and medication costs of three anti-tumour necrosis factor alpha agents in the treatment of rheumatoid arthritis from prospective clinical practice data
Ann Rheum Dis
 , 
2008
, vol. 
67
 (pg. 
1229
-
34
)
30
Arnett
FC
Edworthy
SM
Bloch
DA
, et al.  . 
The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis
Arthritis Rheum
 , 
1988
, vol. 
31
 (pg. 
315
-
24
)
31
Dutch Health Insurance Executive Board
Pharmacotherapeutic compass. Diemen, The Netherlands
2006
32
Wick
MC
Ernestam
S
Lindblad
S
, et al.  . 
Adalimumab (Humira) restores clinical response in patients with secondary loss of efficacy from infliximab (Remicade) or etanercept (Enbrel): results from the STURE registry at Karolinska University Hospital
Scand J Rheumatol
 , 
2005
, vol. 
34
 (pg. 
353
-
8
)
33
Katchamart
W
Johnson
S
Lin
HJ
, et al.  . 
Predictors for remission in rheumatoid arthritis patients: a systematic review
Arthritis Care Res
 , 
2010
, vol. 
62
 (pg. 
1128
-
43
)
34
Hartman
M
Van Ede
A
Severens
JL
, et al.  . 
Economic evaluation of folate supplementation during methotrexate treatment in rheumatoid arthritis
J Rheumatol
 , 
2004
, vol. 
31
 (pg. 
902
-
8
)
35
Edwards
JC
Szczepanski
L
Szechinski
J
, et al.  . 
Efficacy of B-cell-targeted therapy with rituximab in patients with rheumatoid arthritis
N Engl J Med
 , 
2004
, vol. 
350
 (pg. 
2572
-
81
)
36
Briggs
A
Fenn
P
Trying to do better than average: a commentary on ‘statistical inference for cost-effectiveness ratios’
Health Econ
 , 
1997
, vol. 
6
 (pg. 
491
-
5
)
37
Drummond
MF
O’Brien
B
Stoddart
GL
Torrance
GW
Methods for the economic evaluation of health care programmes
 , 
1997
2nd edition
Oxford
Oxford University Press
38
Rutten-van Molken
M
Buscback van
J
Rutten
F
Van kosten tot effecten: een handleiding voor evaluatiestudies in de gezondheidszorg
[Costs and effectiveness: a guideline for studies of evaluation in health care]
 , 
2000
Amsterdam, The Netherlands
Elsevier gezondheidszorg
39
Koopmanschap
MA
Rutten
FF
van Ineveld
BM
van Roijen
L
The friction cost method for measuring indirect costs of disease
J Health Econ
 , 
1995
, vol. 
14
 (pg. 
171
-
89
)
40
Chen
YF
Jobanputra
P
Barton
P
, 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
, vol. 
10
 (pg. 
iii
-
xiii
)
41
Finckh
A
Bansback
N
Liang
MH
Cost-effectiveness of biologics in early rheumatoid arthritis
Ann Intern Med
 , 
2010
, vol. 
152
 (pg. 
333
-
4
)
42
Quinn
MA
Conaghan
PG
O’Connor
PJ
, 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
, vol. 
52
 (pg. 
27
-
35
)
43
Emery
P
Genovese
MC
van Vollenhoven
R
, et al.  . 
Less radiographic progression with adalimumab plus methotrexate versus methotrexate monotherapy across the spectrum of clinical response in early rheumatoid arthritis
J Rheumatol
 , 
2009
, vol. 
36
 (pg. 
1429
-
41
erratum in J Rheumatol 2010;37:2198
44
Verstappen
SM
Jacobs
JW
Hyrich
KL
Effect of anti-tumor necrosis factor on work disability
J Rheumatol
 , 
2007
, vol. 
34
 (pg. 
2126
-
8
)
45
van Vollenhoven
RF
Ernestam
S
Geborek
P
, et al.  . 
Addition of infliximab compared with addition of sulfasalazine and hydroxychloroquine to methotrexate in patients with early rheumatoid arthritis (Swefot trial): 1-year results of a randomised trial
Lancet
 , 
2009
, vol. 
374
 (pg. 
459
-
66
)
46
van Tuyl
LH
Boers
M
Lems
WF
, et al.  . 
Survival, comorbidities and joint damage 11 years after the COBRA combination therapy trial in early rheumatoid arthritis
Ann Rheum Dis
 , 
2010
, vol. 
69
 (pg. 
807
-
12
)
47
Kobelt
G
Eberhardt
K
Jonsson
L
Jonsson
B
Economic consequences of the progression of rheumatoid arthritis in Sweden
Arthritis Rheum
 , 
1999
, vol. 
42
 (pg. 
347
-
56
)
48
Wong
JB
Singh
G
Kavanaugh
A
Estimating the cost-effectiveness of 54 weeks of infliximab for rheumatoid arthritis
Am J Med
 , 
2002
, vol. 
113
 (pg. 
400
-
8
)
49
Lajas
C
Abasolo
L
Bellajdel
B
, et al.  . 
Costs and predictors of costs in rheumatoid arthritis: A prevalence-based study
Arthritis Rheum Arthritis Care Res
 , 
2003
, vol. 
49
 (pg. 
64
-
70
)
50
Barton
P
Jobanputra
P
Wilson
J
, et al.  . 
The use of modelling to evaluate new drugs for patients with a chronic condition: the case of antibodies against tumour necrosis factor in rheumatoid arthritis - introduction
Health Technol Assess
 , 
2004
, vol. 
8
 (pg. 
iii, 1
-
91
)
51
Clark
W
Jobanputra
P
Barton
P
Burls
A
The clinical and cost-effectiveness of anakinra for the treatment of rheumatoid arthritis in adults: a systematic review and economic analysis
Health Technol Assess
 , 
2004
, vol. 
8
 (pg. 
iii
-
x
)
52
Drossaers-Bakker
KW
De Buck
M
Van Zeben
D
, et al.  . 
Long-term course and outcome of functional capacity in rheumatoid arthritis - the effect of disease activity and radiologic damage over time
Arthritis Rheum
 , 
1999
, vol. 
42
 (pg. 
1854
-
60
)
53
Welsing
PM
van Gestel
AM
Swinkels
HL
, et al.  . 
The relationship between disease activity, joint destruction, and functional capacity over the course of rheumatoid arthritis
Arthritis Rheum
 , 
2001
, vol. 
44
 (pg. 
2009
-
17
)
54
Krishnan
E
Sokka
T
Hakkinen
A
, et al.  . 
Normative values for the health assessment questionnaire disability index: benchmarking disability in the general population
Arthritis Rheum
 , 
2004
, vol. 
50
 (pg. 
953
-
60
)
55
van Gestel
AM
Prevoo
ML
van’t Hof
MA
, et al.  . 
Development and validation of the European League Against Rheumatism response criteria for rheumatoid arthritis. Comparison with the preliminary American College of Rheumatology and the World Health Organization/International League Against Rheumatism Criteria
Arthritis Rheum
 , 
1996
, vol. 
39
 (pg. 
34
-
40
)
56
Gabriel
SE
Drummond
M
Maetzel
A
, et al.  . 
OMERACT 6 economics working group report: a proposal for a reference case for economic evaluation in rheumatoid arthritis
J Rheumatol
 , 
2003
, vol. 
30
 (pg. 
886
-
90
)
57
Miettinen
OS
The need for randomization in the study of intended effects
Stat Med
 , 
1983
, vol. 
2
 (pg. 
267
-
71
)
58
van der Kooij
SM
Allaart
CF
Dijkmans
BA
Breedveld
FC
Innovative treatment strategies for patients with rheumatoid arthritis
Curr Opin Rheumatol
 , 
2008
, vol. 
20
 (pg. 
287
-
94
)
59
Combe
B
Landewe
R
Lukas
C
, et al.  . 
EULAR recommendations for the management of early arthritis: report of a task force of the European Standing Committee for International Clinical Studies Including Therapeutics (ESCISIT)
Ann Rheum Dis
 , 
2007
, vol. 
66
 (pg. 
34
-
45
)
60
Keystone
EC
Kavanaugh
AF
Sharp
JT
, et al.  . 
Radiographic, clinical, and functional outcomes of treatment with adalimumab (a human anti-tumor necrosis factor monoclonal antibody) in patients with active rheumatoid arthritis receiving concomitant methotrexate therapy: a randomized, placebo-controlled, 52-week trial
Arthritis Rheum
 , 
2004
, vol. 
50
 (pg. 
1400
-
11
)
61
van der Kooij
SM
Le Cessie
S
Goekoop-Ruiterman
YP
, et al.  . 
Clinical and radiological efficacy of initial versus delayed treatment with infliximab plus methotrexate in patients with early rheumatoid arthritis
Ann Rheum Dis
 , 
2008
, vol. 
8
 (pg. 
iii–iv, ix–x, 1
-
105
)

Comments

0 Comments