RA economic burden

A large number of cost-of-illness studies have been performed during the last decades, and have provided quite different estimates. The main sources of discrepancies were the sample characteristics (size, representativeness, RA characteristics and duration), the health-care system organisation and, more important, the methodology used to calculate the costs.

Recently, a systematic review of the literature was performed and attempted to get more homogeneous estimates of RA costs. The literature search focused on cost-of-illness studies, based on data from observational studies or randomised controlled trials (RCTs), and conducted in Western countries from a societal perspective . Twenty-six RA studies were included, mainly concerning established RA. The integration of all RA costs estimates – expressed in 2006 € – resulted in mean direct costs of 4170 € [IQR 2756–4561] and mean indirect costs of 1441 € [IQR 702–1307] and 8452 € [IQR 4144–11 566] according to the friction cost or the human capital approaches, respectively ( Table 1 ). The difference between the indirect costs estimates was largely due to methodological issues. The friction cost method takes into account short-term sick leave (within the so-called friction period) which is important in early RA; however, it ignores the costs related to long-term sick leave and disability leave, which are likely to occur in established diseases . For this reason, the friction cost method probably underestimates the real impact of RA on productivity loss and indirect costs. The paper also describes the non-health-care direct costs. They represent 2284 € [IQR 628–3092], mainly related to home-care services, patient time and informal care by family and friends.

Three main messages can be drawn from this analysis. First, from a general point of view, the overall direct costs related to RA care are around 4000 €, which is consistent with another non-systematic literature review performed previously . Second, no consensual indirect cost overview is possible. The friction cost method might better reflect the real costs for the society, since it takes into account that the productivity loss of a sick worker can be substituted by an unemployed person after a limited period of time. In the situation of such substitution, there is no productivity loss anymore at the societal level. Conversely, the human capital method might better reflect the consequences of the disease on work capacity, since it ignores the possible substitutions by healthy persons. Other sources of difference in indirect cost estimates can be found in the heterogeneity of the studied populations with regard to job opportunities, employment rate and social support in case of limitations in the workplace. Finally, there is an urgent need for methodological consensus on how resource utilisation and productivity loss should be collected, reported and valued in monetary terms, to enhance comparability between studies and improve the overall clarity of economic evaluation around RA as well as any chronic diseases .

Evolution of costs during the last decade

RA economic burden has evolved during the last decade, mainly due to the launch of biologicals . Although secular trends are difficult to estimate due to the previously mentioned methodological issues, a couple of observations could be made from two observational studies conducted in France, 5 years apart, according to close methodologies. The first study – practis – was conducted in several rheumatology departments in 2000 and aimed to establish the cost of RA just before the launch of biologicals in the market : 1109 patients with established RA were included and their health resource use was determined over a 1-year period. The other one was the Eco-PR study, which was conducted among members of a patient association and collected data in 2005 about RA-related health consumption of 1487 patients with established RA. Twenty-seven percent of patients were treated with TNF-blockers . The comparison of the results of the two studies reveals there was a huge increase in RA economic burden within a 5-year period: 4003 € in practis versus 11,757 € in Eco-PR (2005 actualised Euros) ( Fig. 2 ). Moreover, the cost breakdown dramatically changed over this period with a decrease in hospitalisation costs but at the same time an increase in medication costs due to the biologicals. This observation emphasises the substantial burden of biologicals which have deeply altered the financial burden of RA care; despite this, these agents have been widely accepted in Western countries due to the dramatic efficacy, assuming that part of these costs will outweigh the long-term economic burden of RA by reducing health-care resource use and preventing work disability.

Evolution of RA costs in 3 cross-sectional studies performed according to the same methodology.

The main cost drivers in RA

The average estimates presented above are associated with substantial variability depending on a couple of determinants among which two need to be emphasised.

Functional limitations, disability and disease duration

As it could be expected, the more severe the RA is, the more substantial costs it incurs. Although RA severity is not completely straightforward and consensually defined, disease severity is closely associated with progressively increasing functional limitations, leading to irreversible disability. RA-related disability is strongly correlated to disease activity in the early phase of the disease – reversible damage – and to structural damage, that is, joint destruction, in the later stages– irreversible damage . In other terms, disability is the result of an interaction between disease activity, structural damage and time, that is, disease duration.

The health assessment questionnaire (HAQ) , the most widely used disability measurement tool, is a strong predictor of future pejorative outcomes such as overall morbidity and mortality . Moreover, several economic assessments have also demonstrated that HAQ is a strong cost driver since it is associated with both health-care resource use, that is, direct costs, and at-work or at-home productivity loss, that is, indirect costs . As presented in Fig. 3 , the relation is almost linear with regard to direct costs – in Sweden and United Kingdom a break for people having a HAQ higher than 1 or 1.5, corresponding to moderate disability ( Fig. 3 ). Indirect costs were not taken into account in these costing studies, which have been performed from a payers’ perspective. Since costs of disability pension and early retirement are not covered by the health insurance in these societies they don’t appear anymore in these cost-of-illness studies.

Relation between HAQ and direct and indirect costs based on 3 European studies . C ourtesy of G Kobelt .

Importantly, the role of co-morbidities should be added in the present discussion, with a special attention to cardiovascular diseases, depression and infections. Recent studies show that they contribute significantly to disability and account for substantial costs ; especially in patients with severe and long-standing RA.

Equity considerations: access to care, social security and economic welfare

Substantial differences do exist between countries regarding cost of RA. In the paper by Franke et al. cited above , only cost estimates issued from Western countries were included. However, international comparisons have shown striking discrepancies between Western developed countries and Eastern emerging countries. For example, an European comparison performed in the mid 2000s revealed the total annual RA costs were roughly 15 000 € per patient in Western Europe versus only 3800 € in Eastern Europe .

The relative contributions of the cost breakdown were rather similar between Western and Eastern European countries, but the absolute level of resource utilisation and costs differed ( Fig. 4 ). This reflects differences in organisation of care, frequency of visits to the specialists due to less favourable medical demographics, possibility of inpatient care in regions where hospitals are potentially less numerous and more distant, access to modern therapy such as biologicals and of course health-care insurance coverage . Access to care, and specifically to new and expensive treatment has been shown to be compromised in socioeconomic-deprived populations . In addition, the overall socioeconomic situation in the country, represented by the gross domestic product or the health expenditures for health care of a country, can be a barrier in access to care. On the same line, difference in the organisation of the social security and the prevailing unemployment within a country has been associated with differences in sickness absence and work disability between countries . Interestingly, the value of work and productivity, and consequently the level of indirect costs, is lower in countries with elevated unemployment rates resulting in relatively low indirect costs, despite higher work disability . The above observations lead us to the publication by Sokka et al., based on data collected in 22 different countries, which showed that the average disease activity level as well as of disability clearly related to the economic welfare of this country in terms of gross domestic product (GDP) or expenditures for health care .

Differences in RA costs between Western and Eastern Europe . Courtesy of G Kobelt .

Of course, affordability is an important part of the problem of access to care in the late years. A recent economic report reveals that the proportion of RA patients treated with biologicals differs greatly in the different countries over the world, which has a substantial impact on RA costs . In these patients, RA treatment costs have become quite exclusively related to costs of these drugs – and to hospitalisation related to their administration or surveillance – since the synthetic DMARDs are substantially cheaper than biologicals . This has also been confirmed in two economic evaluations conducted in early RA patients, one based on randomised clinical trial data – the best stragegy (BeSt) trial – and another based on observational data – etude de suivi de polyarthrites indifférenciées récentes (ESPOIR) cohort – . In these two studies, the costs incurred by TNF inhibitors represent a four to sixfold immediate increase of direct costs, biologicals by themselves representing 70–80% of these costs as shown in Fig. 5 . Of course, part of these costs could potentially be offset in future RA care, but it raises the issue of the optimal use of such expensive treatments, that is, the optimal time to introduce them and eventually to taper or stop them. In low-income countries, these balances will be even less favourable since the price of biologicals is relatively higher while the costs for health-care services and value for work productivity are lower.

Immediate but not persistent cost increase due to the use of biologicals in early RA, based on BeSt trial data, adapted from van den Hout et al. In the step-up/step-down therapeutic strategy used in the BeSt trial, infliximab as 1st-line agent (square dots) is associated with substantial increase in direct medical costs compared to methotrexate as 1st-line agent with infliximab introduced eventually in case of inadequate response. However, after 3 years costs difference become smaller since more patients in the initial non-biologicals group need a biological.

Effect of biologicals on health resource utilisation?

The dramatic efficacy of biologicals is associated, as expected, with some reduction in health-resource use by patients suffering from RA . It is important to state that this improvement is not only due to biologicals by themselves but by the striking changes in RA care based on three main paradigms: RA is a medical emergency, start methotrexate (MTX) as soon as possible and target for remission . In this context, biologicals were one of the tools which made possible such a positive evolution.

The impact of biologicals on medical consultations and work-ups is probably real but somehow difficult to assess. RCT data are irrelevant to answer such questions since the patient follow-up is completed in a standardised way with similar assessment visits in the different treatment arms. Observational data are more relevant in this respect, but the potential multiple biases prevent any robust conclusions. Different studies showed that biologicals use was associated with reduction of medical doctors and other health professional visits, as well as work-ups while others showed only marginal changes. In a small UK-based case–control study resource utilisation prior and after starting anti-TNF therapy was evaluated . Outpatient clinic visits increased slightly in patients receiving subcutaneous anti-TNF therapy, but decreased considerably for patients receiving intravenous anti-TNF who were regularly seen in the hospital for infusions. The impact of such a reduction on RA costs remains minimal compared to the overall burden of the disease and its treatment.

The impact of biologicals on surgery deserves more attention since hospitalisation for orthopaedic surgery was a substantial component of RA costs in the pre-biological era. Several studies have reported a trend for reduction in the need for total joint replacement in RA patients diagnosed after 1985–1990 or a shift towards less joint-sacrificing surgical procedures . For example, a study conducted in Finland in 2007 showed that the incidence of total joint replacement was stable in the RA population between 1986 and 2005, although it increased in the general population during the same period . Another study was based on data from the Olmsted County, a region in the United States where the general population medical information has been collected prospectively since 1955 . This work reported a significant reduction in the need for orthopaedic surgery in RA patients diagnosed after 1985 compared to those in the preceding decades ( p < 0.001 after adjustment). Recently, additional data confirmed this trend: in the Japanese IORRA registry, a decrease in the need for different types of orthopaedic surgery was noted as of 2003 when the first biologicals became available in this country ( Fig. 6 ) .

Bi-annual incidence of different types of orthopedic surgery procedures between 2001 and 2007 in the Japanese IORRA registry, adapted from Momohara et al. .

Effect of biologicals on work participation

The impact of RA on work participation is substantial. Work disability occurs as soon as in the early years of the disease . A study in the Netherlands showed that 51% of RA patients experienced impaired productivity in the work place within the first years of the disease, with 25% of them recognised as work disabled . In the ESPOIR cohort mentioned above, 20% of RA patients were on sick leave and 12% on disability leave at 3 years after disease onset . While sick leave might be more important in the first years, work disability increases in the later years, likely because those having the most frequent and longest sick leave become gradually work disabled. Since there is a strong relation between disease activity and functional limitations due to RA on the one hand and work productivity impairment on the other hand, substantial improvement of work participation of RA patients was expected after the launch of biologicals. Notwithstanding, data on worker participation did not seem to have changed greatly in the biological era as was shown in the multinational study QUEST-RA in which rates of probability to continue working in working patients whose symptom had begun during the 2000s were 80% at 2 years and 68% at 5 years . This study showed major impact of RA on work disability. It was especially striking that patients who continued to work in low-GDP countries had a disease activity as high as or higher than patients in high-GDP countries who were work disabled. These discrepancies may partly be explained by possible differences in social security systems, economic situations, job types (manual vs. non-manual jobs) and the access to care including availability of biological therapies. In an observational study by Yelin , employment rates were found to be higher in patients who participated in etanercept clinical trials compared to patients who were not treated with biological agents and participated in an observational study (adjusted probability of employment 0.73 vs. 0.53; difference −0.20 (95%CI –0.32, −0.09). By contrast, in an observational study including patients with established RA and registered with the BSRBR, no association between the use of anti-TNF therapy and employment status 3 years after starting anti-TNF therapy was observed (adjusted odds ratio (OR) for RA control patients vs. RA anti-TNF patients 0.80, 95%CI 0.36, 1.81) . In a nested case–control study, again no association between biological agents and employment status was observed . However, a protective effect was observed in patients with disease duration of less than 11 years when stratifying for disease duration (OR for anti-TNF use 0.5, 95%CI 0.1, 0.9). When interpreting these data, we should realise that these studies report from the early phases of the introduction of biologicals, and in many countries, in which biological agents are only given to patients who fulfil certain criteria for severity and the timing of introducing these drugs may be too late if employment is one of the outcome measures. Many patients may already have severe joint damage and it has been shown that this is associated with irreversible disability and the inability to improve this disability declines with disease duration. Even if patients with long-standing RA who become work disabled respond to biological therapies, the likelihood of regaining employment is very low. The key is therefore to prevent patients from becoming work disabled in the early stages of the disease. In patients with RA who are working at start of biological therapy, response to treatment decreases the likelihood of lost employment and significantly improves productivity at work . When looking at the data on employment in four recent RCTs that studied patients with early RA, small but significant improvements in employment were seen in two of them . To account for the short duration of many studies on employment/work disability, employability has been introduced as an outcome measure and can be applied to those working and currently not working due to health. It includes a subjective evaluation of the individual whether he/she feels able to work. This would adjust for persons who are not employed but feel better due to therapy but are not able to find work (e.g., because no job is available). In a study by Smolen et al. no effect of infliximab on actual employment rates was seen, but an increase in employability was observed in this 1 year follow-up study in patients with early RA . Less is known about employability in patients with long-standing RA treated with biological therapies.

Since sick leave and presenteeism (i.e., at-work productivity loss) in those with paid work often precede work disability, it is important to concentrate on studies exploring the impact of biologicals on these work outcomes. All studies with sick leave as work outcome showed an improvement, either in comparison to sick leave in the control group (in four RCTs and in two controlled cohorts) or compared to sick leave before the start of the biological (in four uncontrolled studies). In one RCT, a short-term beneficial effect on absence from work was not sustained in a long time horizon. In a population-controlled study by Olofsson et al. , a decrease was reported for the relative risk (RR) of being on sick leave after initiation of biologicals when compared to general population data. However, when comparing the RR for sick leave 1 year after treatment with 1 year before treatment, almost no difference was noticed (even a slightly higher level of sick leave in patients), suggesting patients returned to the sick leave rate before the increase of sick leave surrounding the initiation of biologicals. In the same study, no impact was seen on the rates of disability compared to the general population. Difficulties in job performance and/or productivity loss while at work, also referred to as ‘presenteeism’, can be measured by several approaches . Although content validity and reliability of most of these measures are relatively good, correlations between different instruments are low to moderate . The use of these instruments is also appropriate in studies with a short follow-up duration in which change of actual employment is unlikely to occur. It remains unclear whether productivity loss while being on the job as reported by a patient also results in economic productivity loss (and therefore income loss) at the workplace. Many companies will have a so-called elasticity to account for reduced productivity, since work can (i) sometimes (partly) be taken over by others in their own hours or (ii) can be postponed and completed during later working hours by the worker himself or herself . Others argue that if other workers of the team are dependent on the work of one person and there is no possibility for immediate compensation by others, the productivity loss reported by one employer might underestimate the economic productivity loss of the workplace . In four RCTs in patients with early RA, presenteeism improved more in those receiving a biological (combination) than in patients treated with MTX alone. Effects up to 104 weeks were seen. Interestingly, the combination of adalimumab and MTX in early RA, MTX-naïve patients reduced presenteeism significantly compared to MTX alone ( p < 0.05), while adalimumab monotherapy was not able to significantly improve presenteeism compared to MTX monotherapy. In five cohort studies, improvement was seen in presenteeism, but in one of these studies results were conflicting when another approach to assess presenteeism was used and in another study the effect after 6 months was lost after 1 year .