The main finding in this study is an estimated 47.3% reduction (55.6% CPI-adjusted) in the annual per-patient cost of b/tsDMARD from 2010 to 2019 in Norway. During this period, a national tender system for the prescription of b/tsDMARDs was implemented. The estimated annual cost reduction for naïve b/tsDMARD users was 75.4% (79.5% CPI-adjusted). Cost simulation using 2020 tender results on the 2019 population treatment data found that reduction increased further to 81.5% (85.0% CPI-adjusted) from 2010 for naïve patients.
The findings in our study suggest that the implemented tender system for b/tsDMARD procurements in Norway for the last ten years may have facilitated positive competition between pharmaceutical companies and thus served as a market mechanism to reduce prices. The Norwegian Pharmaceutical Procurement Cooperation, a subdivision of the Norwegian Hospital Procurement Trust, has annually released lists of their recommendation for b/tsDMARDs use based on the results of the tender. The prescribing physicians are not obliged by law to follow the annual recommendations and may therefore choose another drug in case of individual reasons. However, the regional health trusts strongly advise and monitor the adherence to the annual (tender-based) recommendations.
The expiration of patents for reference bDMARDs has enabled the development and production of biosimilar bDMARDs, reaching the market at lower costs. In 2014 CT-P13 infliximab was the first biosimilar to reach the Norwegian market, followed by SB4 etanercept in 2016 [12, 13]. In the 2019 Norwegian tender process, several companies manufacturing biosimilar adalimumab drugs gave price offers. However, the reference adalimumab won the tender by offering a lower price than what was offered for the biosimilars. The same was seen for etanercept in 2020, where the reference and not a biosimilar drug won. This shows that biosimilars influence the competition between pharmaceutical companies by influencing producers of reference bDMARDs to reduce their prices in order to win the tender. In 2020 however, the biosimilar GP2017 adalimumab won the tender process.
In Denmark, estimated accumulated price and quantitative data have been published for infliximab, etanercept, and adalimumab after the expiration of a patent [14, 15]. When the adalimumab biosimilar reached Denmark's market in October 2018, the price for adalimumab dropped by 83% within three months. Whereas between September 2018 to September 2019, the use of adalimumab increased by approximately 35% [14].
The third mechanism used in Norway and Denmark to promote rapid cost reduction for bDMARDs is the recommended switch to the cheapest available substance when generics or biosimilars are available.
As shown in our study, the impact of a tender system to reduce drug cost is a mechanism that may increase the availability of b/tsDMARDs to treat inflammatory arthritis, e.g., RA. This may be particularly important for low-income countries where RA patients have been shown to have higher disease activity than higher-income countries [4–6, 16, 17].
The previously documented improvement in clinical outcomes for RA patients in the new millennium in Norway [2, 3] and other countries [19–24] was also found in our study. Aga et al., in the NOR-DMARD multicenter study, found that remission rates in RA patients after six months of TNFi (and methotrexate) treatment had increased from 17% in the period 2000–2002 to 46% in the period 2009–2010 [3]. Disease duration before starting a TNFi had decreased from a median of 8,0 years (2000–2002) to 3.8 years (2009–2010) [3]. In comparison, in our study, the percentage of patients in DAS28 remission increased from 34.8% in 2010 to 61.3% in 2019, whereas disease duration in RA patients who started naïve on b/tsDMARDs did not change substantially.
Treatment with b/tsDMARDs in randomized clinical trials has been shown to improve occupational outcomes [25, 26, 27]. From the Swedish bDMARD registry, 35% of work-disabled RA patients with a disease duration of fewer than five years were found to regain their work ability within three years after starting a TNFi. With a disease duration of five years or more, the work recovery proportion was only 14% [28]. In our study, we did not see a significant change in the proportion of enabled workers.
In the QUEST-RA study with data collected between 2005 and 2009 from 32 countries, 37% of previously work-enabled RA patients aged 65 years and younger reported occupational disability at the onset of RA symptoms (median observation period of 9 years) [29]. Despite the major differences in disease activity in their study, there was no significant difference in the proportion of work-enabled RA patients between countries with high and low gross domestic product (GDP). RA patients in low-GDP countries remained working despite high levels of disability and disease activity, suggesting that cultural and economic differences between societies also impact work disability rates in RA patients [29].
Our study's major strength is that the data collected is standardized for all RA outpatients independent of treatment using the same hospital computer system. This is in contrast to some registry-based studies that either only included selected patient groups using b/tsDMARDs or patients who initiated treatment with csDMARDs and/or b/tsDMARDs (e.g., the Norwegian NOR-DMARD registry) [30]. Another strength is that the included patients come from ten centers spread across Norway. Selection bias, if present, would most likely affect the first years of the ten-year period as the number of registered patients was lower than at the end of the period. However, no significant changes were seen between the RA patients for age, sex, CCP, and RF status.
Furthermore, comparing the estimated mean prevalence for RA of 0.30% in 2019 (single centers range 0.20–0.46%) in our study with a population-based prevalence of 0.44% in Oslo (1994) for the age group 20–80 years and 0.47% in Tromsø (1994) for the age group 20 years and older indicate a low grade of selection bias, at least in some centers [7, 8]. RA patients followed by privately practicing rheumatologists have not been included in the analysis and may partly explain lower prevalence estimates in some centers. However, we have reason to believe that both internal validity for each center and external validity for Norway are satisfactory.
The relatively high rate of missing data for disease activity measures is a limitation. Nevertheless, as argued above, we find this less likely to be caused by a systematic bias and is most likely based on random. Another limitation is the reduced effort of including patients in the BioRheuma projects during the early phase of the ten-year period. Therefore, the increasing percentage of included patients may be strongly affected by the examining physician's interest in including the patient into the GoTreatIt Rheuma database. Also, it cannot be excluded that the improved disease outcome across the ten years may have improved due to other factors such as earlier diagnosis, starting b/tsDMARDs at a lower disease activity, improved self-management, fewer comorbidities, and other aspects that may have reduced the patient global assessment (a key component of DAS28) besides the effect of b/tsDMARDs.