Medical therapy for rheumatoid arthritis—value for money?

The economic burden of rheumatoid arthritis (RA) is considerable and has been well documented [1–4]. The cost of managing the disease approximates to that of treating ischaemic heart disease [1] and several studies have suggested that the direct costs of treating patients with RA are two to three times higher than age‐ and sex‐matched subjects without arthritis [3–5]. In addition, the cost of treating comorbidity is also higher in patients with RA than in non‐RA controls [5]. In England, the total economic impact of RA has been estimated to be £1.256 billion, over half of which is accounted for by loss of earnings [6]. In‐patient and long‐term institutional care account for over 50% of the direct medical cost, while the cost of drugs, including monitoring and toxicity management, accounts for only about 15%. A similar distribution of cost was reported in a recent systematic review of the economics of RA where between 8 and 24% of the direct cost of treating RA was on drugs, monitoring and treating side‐effects [7]. Requirements for daily care, assistive devices and home modifications are three times higher than in non‐arthritic controls [8]. Yet even these data fail to express adequately the overall economic impact of the disease. This is due largely to the fact that the so‐called indirect costs of RA, of which a major component is loss of productivity due to morbidity and premature death, at least equal and, in many studies, far outweigh the direct costs [3, 9]. Loss of employment occurs early in RA, with recent evidence suggesting that many patients give up work prior to their first contact with a rheumatologist and certainly before they have received a disease‐modifying anti‐rheumatic drug (DMARD) [10]. Until recently DMARDs have been inexpensive when compared with new agents in the fields of cardiology, oncology and HIV medicine and also inexpensive compared with the costs of joint replacement surgery. However, conventional DMARDs have modest efficacy, unfavourable toxicity profiles and require frequent monitoring [11]. Rates for remaining on therapy are therefore low and probably only reach about 50% at 5 yr for methotrexate (MTX) [11–14]. This is important as the duration of effective DMARD therapy has been shown to correlate closely with outcome [15]. From an economic perspective, the cost of monitoring and treating side‐effects increases the cost of prescribing a conventional DMARD by 2‐ or 3‐fold compared with the cost of acquiring the drug [16]. It is against this background that the affordability of new biologic agents, which are considerably more expensive than conventional therapy, needs to be examined. Budgetary constraints in the provision of healthcare make rationing inevitable. An informed decision about whether to fund a new treatment requires consideration of what health benefits would be forgone in selecting the new treatment at the expense of competing demands on the healthcare budget. For example, the annual ‘opportunity cost’ of treating one patient with RA with a biologic may be to perform one or two fewer joint replacements; the cost of treating several patients would fund an additional consultant's salary! On the other hand, if such therapies reduce hospital admission rates or the use of joint replacement surgery, or impact favourably on employment stability, there may be a cogent economic argument for their use. In this regard, the median lifetime additional cost, over and above age‐ and sex‐matched subjects, has been estimated to be between $60 000 and $120 000 per patient with RA [17]. It is therefore apparent that more expensive therapy, such as stem cell transplantation (≈$60 000) may be cost‐effective if remission is induced and the downstream costs of RA are reduced [17]. Health economics provides a toolbox for assessing the cost‐effectiveness of therapy for rheumatic diseases [18]. Economic evaluations may either be included within randomized controlled trials (RCTs) or be based on decision analysis models that incorporate clinical data, often derived from meta‐analyses. Both approaches have strengths and weaknesses. Data from RCTs that include an economic analysis are usually more robust but of necessity focus on short‐term outcome in a disease that may span 30 or 40 yr. Such studies may not always be generalizable as they are conducted on selected patients with a high degree of adherence to treatment. Furthermore, some of the resources may be driven by the research protocol and based on the relatively high costs of teaching and research centres. On the other hand, economic models provide a mechanism for extrapolating the results over time and examining the effect of altering different clinical and economic variables. Such manipulation of data is potentially open to bias and has inevitably engendered a degree of scepticism over the validity of economic models. Nevertheless, if the assumptions used are explicit and considered free of commercial influence, such models do provide tentative insights that may then be tested by more rigorous methods [19]. One of the benefits of decision analysis models, such as the Markov model [20], is that the impact of disease progression on cost can be assessed. In this technique, patients are classified into a finite number of states generally defined by disease severity. Disease progression is represented by transition between states and a clinically meaningful timeframe can be incorporated into this. In RA, Markov states have been constructed according to disability level, defined using the modified Health Assessment Questionnaire [21]. Cost and health utility values for each state have been calculated from population and clinical trial data and are widely used in economic models [21]. Despite a substantial number of RCTs of DMARD therapy, few have included a parallel economic analysis and fewer would fulfil the widely accepted quality standards for such studies [22, 23]. The ‘COBRA’ study is exceptional in this regard [24]. Combination therapy, consisting of step down high‐dose steroid together with MTX and sulphasalazine, was compared with sulphasalazine alone in 156 patients with early RA. Clinical, radiographic and functional outcomes were significantly better in the combined therapy group at 28 weeks. This was maintained at 56 weeks for radiographic progression but not for clinical improvement [24]. Direct costs were comprehensively elicited from a societal perspective. The mean total costs per patient were lower for the combination treatment group ($5519 vs $6511, P=0.37) [25]. Protocol and drug monitoring costs were slightly higher in the combination therapy group but this was off set by reduced length of hospital stay and other costs. It was concluded that, within that particular healthcare system, combination therapy was cost‐effective due to enhanced efficacy at lower or equal cost. However, it should not necessarily be assumed that these conclusions are applicable world‐wide and it is difficult to predict longer term consequences from 56‐week data. Unfortunately the indirect costs were not reported. A cost‐effectiveness analysis of cyclosporin has been performed using clinical data from a meta‐analysis of five RCTs [26]. Economic data were modelled over 1 yr from the perspective of both the health service provider and society. The cost‐effectiveness of placebo, azathioprine and penicillamine was compared. Although clearly superior to placebo, cyclosporin was less effective than either of the two active drugs and, as the overall cost of cyclosporin was two to three times that of either active drug, cyclosporin was not cost‐effective. Recent evidence also suggests that cyclosporin is less cost‐effective than triple therapy with hydroxychloroquine, sulphasalazine plus MTX for the treatment of MTX‐resistant RA [27]. Modification of the inflammatory response in RA using biologic agents is unarguably a major therapeutic advance, but the extent to which they fulfil clinical expectations remains to be seen. While recent data on clinical and radiological benefits certainly seem encouraging [28, 29], the price of these agents has limited their availability to those who fulfil stringent criteria [30]. Furthermore, the cost of the biologic itself will always tend to minimize its cost‐effectiveness, when compared with drugs such as MTX, unless far superior efficacy and safety can be demonstrated. This issue has been studied in detail using a decision analysis model in which the relative efficacy, safety and cost of a putative novel biologic were compared with intramuscular gold and MTX [31]. The model simulated the treatment of a cohort of 10 000 patients using each regime over a 6‐month period. Estimates of efficacy for each drug were taken from published data and the cost of the biologic was based on that of similar agents used to treat other conditions. Sensitivity analysis indicated that although the total cost of conventional anti‐rheumatic drugs was driven primarily by the cost of monitoring and treating side‐effects, the primary determinant of the overall cost for the biologic was the acquisition cost of the medication. Even when the efficacy and safety of the biologic were optimized, MTX and intramuscular gold remained more cost‐effective. It was apparent from a further sensitivity analysis that in order to be more cost‐effective than MTX, the biologic had to be highly effective (complete clinical response >60% of patients) and the cost of the biologic plus monitoring had to be less than $6600 per annum. Such data seem unlikely to win the hearts and minds of budget holders or to widen prescribing opportunities. However, there are several caveats to bear in mind. Rheumatologists in the UK are unlikely to use a biologic as primary therapy in MTX‐naive patients. Current guidelines for UK prescribing recommend that biologics should be reserved for patients who have active disease following an adequate trial of two or more DMARDs including MTX [30]. The cost implications of failing to control active RA in patients selected using such criteria may be very different from those in an unselected group as studied in the above model. For instance, they may require hospital admission, which is the single most important determinant of direct cost for RA in the UK [6, 7]. They may also place greater demand on community or long‐term care services. Although the model did make allowances for changes in the indirect costs according to clinical response, no allowance was made for changes in disability level. Current evidence suggests that this is one of the most important determinants of long‐term cost in RA [21, 32]. Finally, simulation over a 6‐month period is a very short timeframe from which to extrapolate long‐term conclusions. For most rheumatologists, MTX has become the gold standard against which to compare new agents [11]. A combination of efficacy and tolerability results in a higher proportion of patients remaining on the drug than most other DMARDs at 5 yr [11–14]. Swedish data suggest that the use of MTX may be cost‐effective in reducing the costs associated with disease progression over a 5‐yr period [21]. Nevertheless, a substantial number of patients respond inadequately to MTX alone and alternative treatments for this group differ widely in efficacy and cost. A cost‐effectiveness analysis has been performed comparing therapeutic options for MTX‐resistant RA, which included etanercept+MTX, etanercept monotherapy, cyclosporin+MTX and triple therapy with hydroxychloroquine, sulphasalazine plus MTX [27]. The American College of Rheumatology 20% response criteria (ACR 20) and a weighted average of proportions of patients achieving ACR 70, ACR 50 and ACR 20 (ACR 70WR) were used as efficacy endpoints. For each treatment, the additional cost per patient over and above the cost of achieving the same outcome using the next least expensive treatment was calculated. As a baseline reference, the cost of achieving ACR 20 ($1100) and ACR 70WR ($1500) in MTX‐naive patients using MTX alone was calculated. For MTX‐resistant patients, neither cyclosporin+MTX nor etanercept monotherapy was cost‐effective in terms of achieving either ACR 20 or ACR 70WR. Triple therapy was only marginally more expensive per ACR 20 ($1500) or ACR 70WR ($3100) than MTX alone in MTX‐naive patients and a reasonable economic argument for its use in resistant RA can therefore be made. The most efficacious option for MTX‐resistant RA was the combination of MTX and etanercept, but this was 38 times more per ACR 20 and 23 times more per ACR 70WR than the baseline cost per response in MTX‐naive patients. Whether a cost of $34 800/ACR 70WR (or $42 600/ACR 20) can be justified is open to debate. Although the timeframe was again only 6 months, the methodology was robust and account was taken of the economic effects of changing disability levels. It has been suggested that core clinical outcome and basic demographic data should be submitted to a UK database for all new patients commencing biologic therapy [30]. This would provide an ideal opportunity to obtain ‘real life’ economic data that might help address some of the uncertainties regarding the impact of biologics on long‐term costs in RA. It would also seem appropriate that future RCTs incorporate uniform core economic data that may permit economic as well as clinical meta‐analyses. In summary, the economic consequences of failing to slow disease progression in RA far outweigh the cost of any currently available therapy for the disease, including biologics. The problem to be resolved is not so much meeting the cost of newer agents, but demonstrating that these drugs influence the biology of the disease to the extent that long‐term direct and indirect costs are reduced. If they do, then these drugs offer good value for money. In the meantime, the struggle to secure funding for biologics continues and rheumatologists remain in the unenviable position of rationing the limited supply. Finally, despite the lack of data on the impact of biologics on the long‐term cost of RA, it must not be forgotten that biologic therapies offer improvements in disease activity, quality of life and radiological progression that are so far unmatched [28, 29]. This, in itself, is a compelling case for their appropriate use.