Article Text

Extended report
Disease activity-guided dose optimisation of adalimumab and etanercept is a cost-effective strategy compared with non-tapering tight control rheumatoid arthritis care: analyses of the DRESS study
  1. Wietske Kievit1,
  2. Noortje van Herwaarden2,
  3. Frank HJ van den Hoogen2,3,
  4. Ronald F van Vollenhoven4,
  5. Johannes WJ Bijlsma5,
  6. Bart JF van den Bemt6,7,
  7. Aatke van der Maas2,
  8. Alfons A den Broeder2
  1. 1Radboud Institute for Health Sciences, Radboud university medical center, Nijmegen, The Netherlands
  2. 2Department of Rheumatology, Sint Maartenskliniek, Nijmegen, The Netherlands
  3. 3Department of Rheumatology, Radboud university medical center, Nijmegen, The Netherlands
  4. 4ClinTRID, Karolinska Institute, Stockholm, Sweden
  5. 5Department of Rheumatology & Immunology, Utrecht University Medical Centre, Utrecht, The Netherlands
  6. 6Department of Pharmacy, Sint Maartenskliniek, Nijmegen, The Netherlands
  7. 7Department of Pharmacy, Radboud university medical center, Nijmegen, The Netherlands
  1. Correspondence to Dr Wietske Kievit, Radboud Institute for Health Sciences, Radboud university medical center, Nijmegen, The Netherlands; wietske.kievit{at}radboudumc.nl

Abstract

Background A disease activity-guided dose optimisation strategy of adalimumab or etanercept (TNFi (tumour necrosis factor inhibitors)) has shown to be non-inferior in maintaining disease control in patients with rheumatoid arthritis (RA) compared with usual care. However, the cost-effectiveness of this strategy is still unknown.

Method This is a preplanned cost-effectiveness analysis of the Dose REduction Strategy of Subcutaneous TNF inhibitors (DRESS) study, a randomised controlled, open-label, non-inferiority trial performed in two Dutch rheumatology outpatient clinics. Patients with low disease activity using TNF inhibitors were included. Total healthcare costs were measured and quality adjusted life years (QALY) were based on EQ5D utility scores. Decremental cost-effectiveness analyses were performed using bootstrap analyses; incremental net monetary benefit (iNMB) was used to express cost-effectiveness.

Results 180 patients were included, and 121 were allocated to the dose optimisation strategy and 59 to control. The dose optimisation strategy resulted in a mean cost saving of −€12 280 (95 percentile −€10 502; −€14 104) per patient per 18 months. There is an 84% chance that the dose optimisation strategy results in a QALY loss with a mean QALY loss of −0.02 (−0.07 to 0.02). The decremental cost-effectiveness ratio (DCER) was €390 493 (€5 085 184; dominant) of savings per QALY lost. The mean iNMB was €10 467 (€6553–€14 037). Sensitivity analyses using 30% and 50% lower prices for TNFi remained cost-effective.

Conclusions Disease activity-guided dose optimisation of TNFi results in considerable cost savings while no relevant loss of quality of life was observed. When the minimal QALY loss is compensated with the upper limit of what society is willing to pay or accept in the Netherlands, the net savings are still high.

Trial registration number NTR3216; Post-results.

  • Anti-TNF
  • Rheumatoid Arthritis
  • Economic Evaluations
  • Outcomes research

Statistics from Altmetric.com

Introduction

Tumour necrosis factor inhibitors (TNFi) are effective in the treatment of rheumatoid arthritis (RA), decreasing disease activity and functional disability and slowing joint damage progression.1 Different TNFi are widely used, with adalimumab and etanercept being the two most frequently used2; they are among the highest selling drugs worldwide.3 However, TNFi are associated with a (dose-dependent) increased risk of infections and non-melanoma skin cancer.4–7 Furthermore, TNFi treatment is costly: approximately €14 000 yearly per patient.8 Improving the usage of TNFi is therefore warranted.

Previous research showed that dose reduction or discontinuation of TNFi without deterioration of disease activity is possible in a relevant proportion of patients.9–11 A promising strategy might be to taper the TNFi until it is stopped, while carefully monitoring the disease, and increase the dose or restart when necessary. The Dose REduction Strategy of Subcutaneous TNF inhibitors (DRESS) study indeed showed that a disease activity-guided dose optimisation strategy of adalimumab or etanercept in RA is non-inferior to standard tight control without tapering with regard to major flaring.12 The strategy resulted in the successful dose reduction or stopping in two-thirds of patients.

Before such a strategy should be widely implemented in daily practice, we need to know about the cost-effectiveness of such a strategy. Although titrating patients to the lowest dose may save medication costs, it may also lead to increased number of patient contacts and consequently increased healthcare costs. Furthermore, flares that occur after dose reduction might be short-lived and easily treated, or could be prolonged, compromising quality of life. However, if the loss in quality of life is compensated with large cost saving, we might consider it a cost-effective intervention. The aim of this study is therefore to relate a difference in costs to difference in quality adjusted life years (QALYs) between a disease activity-guided TNFi dose optimisation strategy and tight control RA care without protocollised tapering from a societal perspective, as a preplanned secondary analysis of the DRESS study.

Methods

Study design and participants

The DRESS study was a pragmatic, open label, randomised, controlled, non-inferiority (NI) trial, stratified by TNFi. The rationale and design, and primary outcomes have been described extensively elsewhere12 ,13 and are summarised here. Consenting patients with RA using adalimumab or etanercept in any stable dose and interval for at least 6 months, with stable low disease activity at two subsequent visits, were enrolled. Low disease activity was determined by rheumatologist and measured using the DAS28-CRP. The study was performed at the Sint Maartenskliniek in Nijmegen and Woerden, the Netherlands, from December 2011 through May 2014, and was approved by the local ethics committee and registered in the Dutch trial register (CMO region Arnhem-Nijmegen; NL37704.091.11, http://www.trialregister.nl;NTR 3216).

Patients were randomly assigned to the dose optimisation or control group stratified for TNFi. Random sized blocks (block size 3–12) were used, and a ratio of dose optimisation versus control condition is 2:1. For more details, see elsewhere.12 ,13

Procedures

Patients allocated to the control group continued a standardised tight control treatment protocol, aimed at maintaining at least low disease activity. Visits were planned once in three months and patients were encouraged to contact the outpatient clinic when experiencing more symptoms. In the dose optimisation group, patients received identical care as the control group, with the addition of a dose reduction advice. This advice consisted of stepwise increasing the interval between injections every 3 months. For adalimumab, the steps were (1) 40 mg every 21 days, (2) 40 mg every 28 days and (3) stop. For etanercept, the steps were: (1) 50 mg every 10 days, (2) 50 mg every 14 days and (3) stop.

In both groups, treatment was changed in case of a disease activity flare. A flare was defined using a validated criterion: as a DAS28-CRP increase >1.2 or a DAS28-CRP increase >0.6 and current DAS28-CRP ≥3.2 compared with baseline DAS28-CRP.14 A regular treatment protocol was followed, including bridging with glucocorticoids or adding or switching conventional synthetic/biologic disease modifying antirheumatic drugs (c/bDMARDs) in the control group. In the dose optimisation group, the last effective interval was reinstated. If despite this the flare persisted, TNFi was increased until the shortest registered interval, thereafter treatment was switched following the same escalation protocol as the control group. Bridging with glucocorticoids was also allowed.

Outcomes

For this cost-effectiveness analysis, we report on quality of life, measured with the EuroQol-5D5L (EQ5D-5L)15 and healthcare usage with related costs from a medical and societal perspective. Based on the EQ5D-5L questionnaire answers, utilities were calculated using the Dutch tariffs for the EQ5D-3L16 and the Dutch crosswalk value set to converse EQ5D-3L utilities to EQ5D-5L utilities.17 To derive QALYs area under the curve was calculated using the trapezium rule; using this method missing EQ5D values were linear interpolated between available time points on patient level.

Volumes of care were determined in both dose optimisation and control group. Consultations, including telephonic consultations, with the rheumatologist or rheumatology nurse, travelling distance and hospital admissions related to RA were collected. Medication use was registered based on the medical chart, and RA-related absence from work was administered with a patient questionnaire. Because most resource usage was based on electronic record, there were no missing values to be considered. The cost prices for common items like outpatient visits were based on the Dutch Guideline for Cost Analyses.18 Cost prices for medication were retrieved from the Dutch national tariff list.8 For detailed information on the prices used, we would like to refer to the online supplementary appendix. For the valuation of absenteeism from work, we used the friction cost method (3 months friction period). The base year for all prices was 2014. Prices retrieved from other years were converted to 2014 using the general Dutch price index rate. Costs were not discounted, because of the short time horizon of 18 months.

Cost-effectiveness analysis

The cost-effectiveness analysis is following a superiority design; and analyses were done on an intention-to-treat basis. Differences in costs and QALYs between intervention and control group (reference) were analysed for the study period of 18 months. A decremental cost-effectiveness ratio (DCER) was calculated by dividing the cost savings by the QALY loss related to the optimisation strategy. Ninety-five per cent uncertainty boundaries in the DCER were determined non-parametrically using bootstrapping (1000 replications). The incremental Net Monetary Benefit (iNMB) per patient for the dose optimisation strategy compared with the usual care strategy was calculated for varying levels of willingness to accept (WTA) in euro's per QALY lost, using the formula: WTA*incremental QALYs – incremental costs. This results in the net amount of money saved, when the possible loss of QALY is corrected for a certain amount of money; this can then be done for different WTA levels per QALY lost.19

As we wanted to anticipate to lower TNFi drug costs in the future due to the introduction of biosimilars and other market effects on drug pricing, we performed a sensitivity analyses with two different drug prices, namely −30% and −50% of the current TNFi-related costs.

Results

Patients

A total of 180 patients were enrolled, 121 patients in the optimisation group and 59 patients in the control group (figure 1). Baseline characteristics were similar between the two groups, except for higher prevalence of DMARD co-medication in the control group (table 1). Almost no data were missing; 2% of the planned visits and 3–7% missing per baseline variable, thus multiple imputation was deemed unnecessary. There were no differences in missing levels by randomisation group.

Table 1

Baseline patient characteristics

Figure 1

(A) Results of the 1000 bootstrapped replications, presented in cost-effectiveness planes. It graphically presents the uncertainty around the cost-effectiveness ratio. (B) Mean incremental net monetary benefit on the y-axis with the lower and upper limit of 95% percentile, plotted against different levels of euro per quality adjusted life years (QALY) that could be chosen as willingness to accept (WTA) on the x-axis. iNMB, incremental net monetary benefit.

Healthcare usage

Medication use is summarised in table 2. At 18 months, in the dose optimisation group TNFi was successfully discontinued in 20% (95% CI 13% to 28%) and tapered in 43% (95% CI 34% to 53%) patients, whereas in 37% (95% CI 28% to 46%) patients no dose reduction was possible. There was no difference between adalimumab or etanercept concerning the percentages of patients in whom no dose reduction was possible, 38% and 36%, respectively. The mean cumulative TNFi dose over the whole study period in the dose optimisation group was 45% (median with IQR 27–65%) of the defined daily dose (DDD, etanercept 50 mg/week and adalimumab 40 mg per 2 weeks). In the control group, 7% (95% CI 2% to 17%) discontinued TNFi (all because of adverse effects), 8% (95% CI 3% to 19%) tapered TNFi because of low disease activity and 85% (95% CI 73% to 92%) did not dose reduce. The cumulative TNFi dose over 18 months expressed as % of DDD was 100% in most of the control patients, namely in 76% (n=41/54, five missing for cumulative dose).

Table 2

TNF and co-medication use

Concerning co-medication, there were some differences. Intramuscular and/or intra-articular glucocorticoid injections were given in 35% (95% CI 27% to 45%) in the dose optimisation group and in 24% (95% CI 14% to 37%) in the control group (p value 0.26). DMARDs were reduced or discontinued more often in the control group compared with the dose optimisation group, 27% (95% CI 17% to 40%) vs 10% (95% CI 5% to 17%) (p<0.01), while DMARD initiation or dose escalation occurred more often in the dose optimisation group compared with the control group, 13% (95% CI 8% to 21%) vs 3% (95% CI 6% to 13%), p<0.05. At 18 months, the percentage of patients using a DMARD remained lower for the dose optimisation group (61%, 95% CI 52% to 70%) compared with the control group (69%, 95% CI 56% to 80%) (p value 0.61).

Concerning other healthcare usage, there were more (p<0.001) outpatient visits in the dose optimisation group (8.3 visits per patients) compared with the control group (7.3 visits per patient). Additionally, there was no difference in RA-related absence from work, which occurred in 2 of 59 (3%, 95% CI 1% to 13%) and 6 of 121 (5%, 95% CI 2% to 11%) patients in the control and optimisation group, resulting in 7 and 25 days work lost, respectively (ns).

Cost-effectiveness

Results from the 1000 bootstrapped replications concerning mean QALYs and costs for 18-month study period are presented in table 3 and figure 1A. In 84% of the replications, the dose optimisation strategy resulted in a QALY loss (lower left quadrant in figure 1A) with a mean lose of −0.02 QALY (95 percentile −0.07 to 0.02). A 100% of the 1000 bootstrapped replication resulting in cost savings; with a mean cost saving of €12 280 (95 percentile −€10 502; −€14 104). The DCER was €390 493 (€5 085 184; dominant) of savings per QALY lost. The iNMB of dose optimisation is shown in figure 1B. When using a WTA level of €80 000 per QALY, often considered the upper limit the society is willing to accept per QALY lost, the mean iNMB is €10 467 (95 percentile €6553–€14 037) per patient per 18 months. Figure 1B shows that the iNMB decreases with higher WTA value; but up to a WTA value of €175 000 per QALY lost, the lower 95% limit of the iNMB is positive, indicating net cost savings.

Table 3

Mean QALYs and costs for both strategies as a result of 1000 bootstrapped replication

Sensitivity analyses with lower TNFi drug prices showed that the iNMB decreases with lower drug prices. With 30% lower drug prices, the mean iNMB with WTA of €80 000 per QALY would be €4502 (€1017–€7996) and with 50% price reduction, the mean iNMB would be €2737 (−€757–€6101) per patient (figure 2).

Figure 2

Mean incremental net monetary benefit with lower and upper limits of 95% percentile for sensitivity analysis with (A) 30% of drug prices and (B) 50% of drug prices. iNMB, incremental net monetary benefit; WTA, willingness to accept.

Discussion

These cost-effectiveness analyses of the DRESS study show that disease activity-guided dose optimisation of adalimumab and etanercept in RA is non-inferior to tight control care without tapering with regard to disease control, and also results in substantial cost savings. The QALYs were not statistically different between the two strategies but there is an 85% chance that this dose reduction would result in a small (mean difference of 0.02) but not clinically meaningful QALY loss. Furthermore, when we would accept a loss of one QALY when compensated with an amount of €80 000, the net monetary benefit would be around €10 000 per patient per 18 months. Even with expected lower TNFi drug prices in the future, disease activity-guided dose optimisation would still be a cost-effective approach.

To our knowledge, this is the first empirical study reporting on the cost-effectiveness of a dose optimisation strategy of TNFi treatment in patients with RA. In a recent Cochrane systematic review11 on downtitration and discontinuation strategies of TNFi in RA, it was concluded that none of the included sponsor-driven studies assessed costs or cost-effectiveness, although these factors are one of the reasons why clinicians consider lowering the dose or stopping the administration of anti-TNF agents. This highlights the relevance of this study.

A strength of the DRESS study is the high internal validity by the randomised design, the use of validated outcome measures and good data integrity. Additionally, the study was designed as a pragmatic trial with no blinding, no strict inclusion or exclusion criteria and no restrictions on cotreatments. This means that the context and the patients are a good reflection of daily clinical practice, which is especially important for the generalisability of cost-effectiveness results.

A limitation of this study is the fact that despite randomisation, there are some differences in baseline characteristics and treatment during follow-up between the dose optimisation and control group. Patients in the dose optimisation group used less co-medication than patients in the control group. Furthermore, the use of DMARDs and oral glucocorticoids at study end was still more prevalent in the control group. Finally, some patients in the control group also reduced or stopped their TNFi treatment. All three differences can result in biased cost-effectiveness ratio estimates; however, the direction of this bias would lead to underestimation rather than overestimation of the DCER. We therefore think that these between-group differences do not invalidate our conclusions of superiority in cost-effectiveness.

Another limitation of this cost-effectiveness analysis is the relatively short time horizon of 18 months. Effects related to dose optimisation or continuation (control) that may occur after that cannot be estimated. For example, the risks of non-melanoma skin cancers or serious infections might be lower in the dose optimisation group, but within this limited study sample this was not seen. However, when we would have been able to incorporate those effects, it would favour the dose optimisation strategy.

Another long-term effect that could influence cost-effectiveness after 18 months is the fact that 5% of optimisation group versus 0% of control group showed minimal radiological progression (more than the smallest detectable change). If it has been caused by the one-time short-lived flares that were more frequent in the dose optimisation arm, progression will be similar between both groups in following years, and it will not be influencing cost-effectiveness. If it has, however, been caused by lower TNFi use itself, the between-group differences could become clinically relevant after say a decade. One could hypothesise that costs will rise because, for instance, orthopaedic surgeries are necessary or patients become functionally impaired and have further loss of work productivity. This would decrease the net cost savings in the future. A long-term extension study is performed to answer these questions.

The generalisability of results is important point to consider. First, the implementation of tight control as a key factor in dose optimisation strategy could in other healthcare systems and countries be more difficult than in our setting, for example, due to large travel distance for patients.20 ,21 This could result in higher costs of performing a dose optimisation strategy in those situations. But we think that these costs will always be outweighed by the cost savings related to implementing the strategy. Another aspect of generalisability is related to the drug prices of TNFi, which are likely to decrease in coming years. Therefore, we performed sensitivity analyses with a 30% and 50% reduction of the TNFi price to see how robust the conclusions are. We can conclude that dose optimisation would still be superior and result in large net savings, except for the most extreme scenario combining 50% price reduction, and €80 000 WTA per QALY lost, where results are borderline non significant.

In a broader view at generalisability, one also needs to consider the upper limit of what is society willing to accept for a QALY lost. In the Netherlands, this is considered to be €80 000 although not officially stated or documented. With higher values of WTA, the net monetary benefit will decline. Furthermore, it is important to note that it is under debate whether or not the willingness to pay value is the same as the WTA value. It is conceivable that one wants a QALY loss to be compensated with higher amounts than one is willing to pay to gain a QALY. It is even conceivable that one is not willing to accept any loss of QALYs. In the end, it is up to the society and decision makers to decide what one is willing to pay or accept, but with WTA values up to €175 000 cost-effectiveness will still be significant at current drug price levels.

When implementing this strategy, one needs to consider that implementation itself also cost money and resources. However, this effect in real life will be counterbalanced by the cost savings after 18 months, as they will most likely be even higher. This is to be expected, because the majority of extra effort and visits was done in the dose optimisation phase in the first months after initiation of dose optimisation strategy. Indeed, this is corroborated by a study of van der Maas et al who showed a gradual decline of total costs over 1 year in 3-month periods.22

In conclusion, we demonstrated that disease activity-guided dose optimisation results in considerable cost savings while no relevant loss of quality of life was observed. When the minimal QALY loss is compensated with the upper limit of what society is willing to pay or accept in the Netherlands, the net savings are still high. Therefore, implementation of this strategy would improve the cost-effectiveness of RA treatment. Future research should include longer follow-up studies confirming persistence of cost-effectiveness, and implementation studies assessing the generalisability and cost-effectiveness from an implementation point of view.

Acknowledgments

We first express our thanks to all the patients who were willing to participate in this study. We also thank the rheumatologists and other healthcare professionals in the Sint Maartenskliniek Nijmegen and Woerden for participation in patient recruitment and data collection. We are indebted to Chantal Bouman and Nienke Lesuis for help during study inclusion, participating in data collection and scoring radiographs (CB). Our gratitude extends also to Els van den Ende and Alexander Rennings for their role in the data safety monitoring board and Nienke Cuperus for her role as monitor of this study.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

  • Lay summary

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Handling editor Tore K Kvien

  • Contributors NvH, AAdB, AvdM, BJFvdB, FHJvdH, RFvV and JWJB were involved in the study design. NvH, AAdB, AvdM and FHJvdH were involved in the data collection. WK, NvH and AAdB performed the data analyses. All authors were involved in writing, revising and final approving of the manuscript.

  • Competing interests All authors have completed the ICMJE uniform disclosure form at http://www.icmje.org/coi_disclosure.pdf and declare: JWJB received grants and personal fees from Pfizer and AbbVie, during the conduct of the study; grants and personal fees from Roche, BMS, UCB, outside the submitted work. RFvV received grants from AbbVie, BMS, GSK, Pfizer, Roche, UCB and personal fees from AbbVie, Biotest, BMS, GSK, Janssen, Lilly, Merck, Pfizer, Roche, UCB, Vertex, outside the submitted work.

  • Ethics approval CMO region Arnhem-Nijmegen.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement The authors commit to making the relevant anonymised patient level data available on reasonable request.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.