RHEUMATOLOGY

Rheumatology 2011;50:196203 doi:10.1093/rheumatology/keq325 Advance Access publication 14 November 2010

Original article
Long-term effectiveness and safety of TNF-blocking agents in daily clinical practice: results from the Dutch Rheumatoid Arthritis Monitoring register
Wietske Kievit1, Jaap Fransen1, Eddy M. M. Adang2, Alfons A. den Broeder3, Hein J. Bernelot Moens4, Henk Visser5, Mart A. F. van de Laar6 and Piet L. C. M. van Riel1
Abstract
Objectives. Experience with anti-TNF agents for a decade can be used to research the safety and effectiveness of anti-TNF agents in the long term. The objective of this article is to describe drug survival, disease activity, daily functioning, quality of life and adverse events of TNF-blocking agents in daily clinical practice after 5 years of follow-up. Methods. Data from the Dutch Rheumatoid Arthritis Monitoring (DREAM) register of 1560 RA patients were used for analyses (5-year follow-up, n = 174). Drug survival and time to first serious infection or malignancy were analysed by KaplanMeier analysis. Several outcome measures at several follow-up moments were analysed per intention to treat and per protocol. Results. The 5-year drug survival of the first anti-TNF was 45%, and 60% for total use of TNF-blocking agents. Baseline 28-joint DAS (DAS-28) was 5.1 (S.D. 1.3). After 5 years, the mean DAS-28 was 3.2 (S.D. 1.3) in all patients who had started with TNF-blocking agents and 2.9 (S.D. 1.1) in patients who were still on TNF-blocking agents. In the latter group, the HAQ score was 0.88 (S.D. 0.7) and the EuroQol five dimensions (EQ-5D) utility score was 0.7 (S.D. 0.2). Incidence rates of serious infections and malignancies were 2.9 and 0.6 per 100 patient-years, respectively.

Downloaded from http://rheumatology.oxfordjournals.org/ by guest on September 21, 2012

CLINICAL SCIENCE

Conclusion. Five-year follow-up of RA patients treated with TNF-blocking agents showed a 60% drug survival accompanied by sustained low disease activity, normalized function and quality of life similar to that in the general population. The benefit to risk ratio for long-term TNF-blocking therapy remains favourable.
Key words: TNF-blocking agents, Effectiveness, Adverse events. Rheumatoid arthritis, Observational, Biologic register, Long-term,

Introduction
Since the introduction of TNF-blocking agents in the late 1990s the treatment of RA has changed tremendously.
1 Department of Rheumatic Diseases (470), 2Department of Epidemiology, Biostatistics and Health Technology Assessment, Radboud University Nijmegen Medical Centre, 3Rheumatology, Sint Maartenskliniek, Nijmegen, 4Rheumatology, Ziekenhuis Groep Twente, Hengelo, 5Rheumatology, Rijnstate Hospital, Arnhem and 6Arthritis Centre Twente, Medisch Spectrum Twente and University Twente, Enschede, The Netherlands.

Submitted 1 April 2010; revised version accepted 30 August 2010. Correspondence to: Wietske Kievit, Department of Rheumatic Diseases (470), Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands. E-mail: w.kievit@reuma.umcn.nl

There is abundant evidence from randomized controlled trials (RCTs) that TNF-blocking agents are efficacious and safe [16] but data on their long-term intended and unintended effects remain relatively scarce. The RCT, as an experimental design, is the gold standard in proving efficacy and safety of a new drug because of its high internal validity. However, external validity is regularly hampered by the controlled setting of an RCT with stringent inclusion and exclusion criteria, limiting generalizability to daily clinical practice [7]. It has been shown that the efficacy of TNF-blocking agents in RCTs exceeded the effects of these drugs in clinical practice, which is mainly caused by the stringent inclusion criteria [8, 9] as well as control for adherence in trials. Furthermore, the follow-up time in RCTs is relatively

! The Author 2010. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Five-year follow-up on TNF blockers from the DREAM register

short and open-label extensions of RCTs include a positive selection of patients originating from the trials. Longer follow-up of TNF-blocking agents is especially necessary for the monitoring of infections and malignancies caused by long-term immune suppression. For the study of adverse events, randomization is generally not necessary, as confounding by contraindication, as a rule, hardly occurs [10] and then an observational study is an appropriate design. Furthermore, the limitations of the external validity of RCT data lead to questioning of their usage for reimbursement decisions [11, 12]. As a consequence of these needs, biologics registers were established in several European countries in which unselected RA patients from routine care treated with biologic agents are followed [13]. In The Netherlands, the Dutch health care authority funded the start of what is now called the Dutch Rheumatoid Arthritis Monitoring (DREAM) register [14]. The primary aim of the register is to monitor and evaluate the safety and effectiveness of TNF-blocking treatments for RA in daily clinical practice. Currently, monitoring of other treatment strategies like other biologic agents and remission-induction strategies are also goals of the DREAM register. The DREAM register is unique compared with other European registers because many diseaserelated and quality of life outcomes are assessed at visits with regular time intervals even in the long term. Therefore, it is possible to report on effectiveness as well as safety using the data from the DREAM register. The aim of this article is to describe the procedures and outcome assessment of the DREAM register and to report on the primary aim of the DREAM register: the cohort of patients on TNF-blocking agents. We provide data on the 5-year drug survival and 5-year outcomes on disease activity, daily functioning, quality of life and serious adverse events of TNF-blocking agents in daily clinical practice.

DMARDs including MTX and absence of an absolute contraindication for anti-TNF (e.g. pregnancy, presence of a serious infection) are required. Patients starting for the first time on TNF-blocking agents are included.

Treatment
Although dosing of anti-TNF and all other medications was at the discretion of the attending rheumatologist, generally patients have started treatment with the Dutchlabelled doses. All biologic agents were usually prescribed in combination with MTX; however, any other DMARD, combination of DMARDs and CSs can be prescribed. All start and stop dates, doses, changes in dose and reasons for changes have been registered for all anti-TNF, DMARDs and CSs.

Data management
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on September 21, 2012

Data collection and management was performed using optical readable Case Record Forms that were centrally read at the Radboud University Nijmegen Medical Centre. Periodically, the database is checked for inconsistencies, which are corrected using source documents. All participating centres are visited half-yearly by the study coordinator for data monitoring. In order to minimize inter- and intra-observer variability, all research nurses who perform the joint counts for the clinical assessments are half-yearly trained by experienced teachers.

Outcome measures
Clinical assessments are performed by trained research nurses, at baseline and every 3 months for the first 2 years and every 6 months thereafter. The following assessments are performed: the 28 tender- and swollen-joint counts, ESR, patients visual analogue scales for general health and for pain, and the Dutch version of the HAQ Disability Index (HAQ-DI) [16, 17]. Disease activity was calculated from clinical assessments using the DAS-28 [18] and response was calculated using the EULAR response criteria [19]. In pre-defined subsamples (n = 830) of patients, quality of life was assessed using the EuroQol five dimensions (EQ-5D) [20] and the Short-Form 36 (SF-36) [21]. The eight domains of the SF-36 were combined into two summary measures, the physical and the mental component scales [22]. On the basis of the answers on the five items of the EQ-5D, a utility (value ranging from 0 to 1) was calculated [20]. During each visit, patients are asked if any adverse events have occurred during the past 3 months. Levels of adverse events are categorized as minor, mild, major or life threatening. For the current analysis, major and life-threatening adverse events were combined in order to fulfil the definition of a serious adverse event by the US Food and Drug Association that states: events disabling daily activities persistently or significantly, needing hospitalization or being life threatening.

Methods
Design
The study design is a multi-centre register on the use of biologic agents in daily clinical practice. Since February 2003, all RA patients who started for the first time on one of the biologic agents in one of 13 centres in The Netherlands have been included in the DREAM register. Patients in this observational study are not randomly assigned to a specific treatment. Data collection is continued even when patients have stopped using biologic agents. Written informed consent is obtained after the patient has been informed orally and after getting written patient information from their attending rheumatologist.

Inclusion criteria
Inclusion criteria for the register are in line with the Dutch regulations for reimbursement of TNF-blocking agents. All patients have a diagnosis of RA according to the 1987 ACR classification criteria [15] and a 28-joint DAS (DAS-28) >3.2. Prior treatment with at least two other

Analyses
Baseline person and disease characteristics were examined for each of the three anti-TNF agents (adalimumab,

www.rheumatology.oxfordjournals.org

197

Wietske Kievit et al.

etanercept, infliximab) separately. For these analyses, values are expressed as mean (S.D.), median (inter-quartile range) or n (%), as appropriate. Drug survival of the biologic agents was analysed with the KaplanMeier technique in which patients who were still using the drug at the time of analysis were censored. A temporary interruption of treatment of <3 months was ignored. The reason to stop treatment and medication use after discontinuation of the initial biologic agent was analysed using frequency tables. Possible factors related to TNF survival (age, gender, RF, presence of erosions, previous DMARDs, disease duration, MTX co-medication, DAS-28 at baseline) were explored using a Cox regression with backward stepwise selection. The criterion for variable removal from the model was a significance level of 0.05. The number of patients with dose increases and dose decreases as well as the number of patients with MTX or oral prednisone were described using frequency tables. Data on effectiveness [DAS-28, HAQ, EQ-5D utility score and the physical component scale (PCS) of the SF-36] were analysed on both an intention-to-treat (ITT) basis and a per protocol (PP) basis. For the ITT analyses, all patients were analysed in the group of medication on which they initially started, regardless of whether they received or adhered to that treatment for the full analysed period. It was possible to perform such an ITT because data collection continued after patients had stopped using their initial TNF-blocking agent. PP analyses were performed for patients who continued their first TNFblocking agents and for patients who continued treatment with any of the TNF-blocking agents, including switching. Time to first serious infection or malignancy was graphically analysed with the KaplanMeier technique in which patients who did not have one of these events were censored at the date of data retrieving. Incidence rates for serious infections and malignancies were calculated by dividing the number of events by the number of patientyears. Only serious infections within the treatment period of the first TNF-blocking agent and the treatment time of the first agent were used for analysis.

start of the initial biologic agent and a higher HAQ at baseline were predictive for exclusion from the study. Table 1 shows the baseline characteristics and the mean follow-up time of patients starting on one of the biologic agents (n = 1560). The mean follow-up time ranged from 33 to 40 months and a total of 174 patients had at least 5-year follow-up. Patients starting with one of the TNF-blocking agents had median disease durations of 5.56.2 years, while nearly half of them had high levels of disease activity according to a DAS-28 > 5.1 (Table 1). Before the start of TNF-blocking agents, all patients used at least two DMARDs, mostly MTX (97%). SSZ (74%), HCQ (49%) and oral prednisone (42%) were also frequently used before starting TNF-blocking agents.

Treatment
Adalimumab was started with the labelled dose in 96% of the cases. Etanercept was started using the labelled dose in 95% of the cases; in 53%, this was 25 mg twice weekly and in 42% this was 50 mg once per week. Infliximab was started in 14% of the cases with 4 mg/kg/8 weeks, which is a dose that is higher than recommended, and 2% started with even higher doses. All patients who started with the 4 mg/kg dose of infliximab did so because the hospital standards indicated the whole vial should be used if the patient needs >220 mg in order to prevent wastage. Within 5 years, 81 (13%) patients on adalimumab, 38 (5.4%) patients on etanercept and 68 (27%) patients on infliximab had a dose increase, whereas 4 (0.6%) patients on adalimumab, 10 (1.4%) patients on etanercept and 7 (2.8%) patients on infliximab had a dose decrease. In total, 1112 (71%) of the patients used MTX and 529 (34%) used CSs at baseline or it was added to the treatment within 3 months after the start.
Downloaded from http://rheumatology.oxfordjournals.org/ by guest on September 21, 2012

Drug survival
The 5-year drug survival (solid line, Fig. 1) of the first TNF-blocking agents was 45%, with the highest discontinuation rate in the first year. The cumulative drug survival was 70% after 1 year, 62% after 2 years, 55% after 3 years, 47% after 4 years and 45% after 5 years. In total, 694 patients stopped their first anti-TNF in the course of 5 years. The reason for stopping the first TNFblocking agent was inefficacy in 282 (41%) and adverse events in 243 (35%) out of 694 patients who stopped their first TNF-blocking agent. Patients regularly [353 (51%) of 694 patients)] started with another TNF-blocking agent after discontinuation of the initial TNF-blocking agent. However, a re-challenge of the same TNF-blocking agent was performed in 50 (7%) of the 694 patients. An additional group of patients switched to one of the conventional DMARDs [96 (13%) of 694 patients)] or did not switch but continued their DMARD [169 (24%) of 694 patients)]. The drug survival of TNF blockers in total, including switching to another TNF-blocking agent, was 60% (dotted line, Fig. 1), according to the KaplanMeier analyses. Of all patients still on TNF-blocking agents at the time of the analysis, 78% of patients were on their first TNF-blocking agent, 22% were on their second (19%) or

Results
Patients
By December 2009, 1560 patients had been included in the DREAM biologic register. From this cohort, 264 (17%) patients have been withdrawn after inclusion for several reasons: 138 (52%) patients refused further participation; 49 (19%) patients were excluded because their hospital no longer participated in the DREAM register; 29 (11%) patients were no longer able to participate due to severe comorbidities; 24 (9%) patients died; 20 (8%) patients preferred to be treated in another (non-DREAM) hospital; and in 2 (1%) patients, the diagnosis of RA could not be confirmed. The data of these patients, except for the last two patients, are included in the analyses until dropout. A multivariate analysis of all baseline variables showed that higher age, a larger number of DMARDs previous to the

198

www.rheumatology.oxfordjournals.org

Five-year follow-up on TNF blockers from the DREAM register

TABLE 1 Baseline characteristics

Characteristic Follow-up time, mean (range), months Female, n (%) Age, mean (S.D.), years Disease duration, P50 (P25P75), years RF positive, n (%) Previous DMARDs, P50 (P25P75) DAS-28 at start, mean (S.D.) HAQ at start, mean (S.D.) Adalimumab, n = 606 35 (078) 423 (70) 54 (12) 6.1 (2.512.7) 434 (74) 3 (24) 5.0 (1.3) 1.3 (0.9) Etanercept, n = 700 33 (079) 483 (69) 55 (15) 5.5 (2.112.2) 486 (71) 3 (24) 5.2 (1.3) 1.3 (0.7) Infliximab, n = 254 40 (282) 179 (71) 57 (13) 6.2 (2.412.3) 192 (77) 3 (24) 5.1 (1.2) 1.4 (1.0)

P50: median value; P25: lower interquartile range; P75: upper interquartile range.

FIG. 1 Drug survival of the first TNF-blocking agents and TNF-blocking agents in general (including switching).
1.0 0.9 0.8

Drug survival

0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0 1 2 3 4 5 First anti-TNF Anti-TNF use

TNF-blocking agent (PP first anti-TNF use); the remission rates (DAS-28 < 2.6) were 43% (50/117) and 48% (38/79), respectively. In all patients who had started with TNFblocking agents (ITT analysis), the HAQ score decreased from 1.3 (S.D. 0.7) to 0.96 (S.D. 0.7). In patients who were still on their first TNF-blocking agent after 5 years (PP first anti-TNF), the HAQ decreased to a value of 0.88 (S.D. 0.7). Quality of life, measured with the EQ-5D and the PCS of the SF-36, improved. The EQ-5D utility score increased from 0.56 (S.D. 0.3) to 0.70 (S.D. 0.2) according to the ITT analysis and increased to 0.75 (S.D. 0.2) in patients who are still on their first TNF blocker at 5 year follow-up. The PCS increased from 30 (S.D. 9) to 36 (S.D. 12) according to the ITT analysis and to a value of 38 in patients still on their first TNF blocker at 5-year follow-up.

Downloaded from http://rheumatology.oxfordjournals.org/ by guest on September 21, 2012

Survival probability

Time, years

Adverse events
Figure 3 shows the time to first serious infections (solid line) and the time to first malignancy (dotted line) within 5 years of follow-up. In 28 (1.8%) patients, a malignancy was reported during 5017 patient-years of follow-up resulting in an incidence rate of 5.6/1000 patient-years. Of the malignancies, there were six gastrointestinal cancers, four non-melanoma and two melanoma skin cancers, four lymphomas, three prostate cancers, two female breast carcinomas and seven other cancers. In 81 (5.2%) patients, at least one serious infection was reported during 3067 patient-years of follow-up (the usage time of the first TNF-blocking agent), resulting in an incidence rate of 2.6/100 patient-years. The infections were 34 upper respiratory tract infections, 24 skin infections, 11 urological infections, 8 joint infections, 3 gastrointestinal infections and 1 opportunistic infection. In 18 patients, two or more infections (with a maximum of four) were reported.

third (3%). Of all patients who were no longer on TNF-blocking agents, 15% started on one of the other biologic agents, 26% started with a new DMARD, 12% had a re-challenge of any of their TNF-blocking agents and 47% did not switch but continued their current DMARD. Factors related to discontinuation of the first anti-TNF were the presence of erosions [HR 0.74 (0.62, 0.89)] and a higher number of previous DMARDs [hazard ratio (HR) 1.12 (1.06, 1.18)]. Factors related to discontinuation of any anti-TNF were the presence of erosions [HR 0.70 (0.56, 0.88)] and a higher age (in categories of 10 years) [HR 1.14 (1.05, 1.24)].

Effectiveness
Figure 2 shows the effectiveness of TNF-blocking agents considering disease activity, functional disability and quality of life. After 5 years, the mean DAS-28 in the groups of patients who had started with one of the three TNFblocking agents (ITT analysis) was 3.2 (S.D. 1.3; moderate disease activity). Of the 174 patients, 66 (38%) had a DAS-28 < 2.6, which is regarded as near-remission. The mean DAS-28 was low [2.9 (S.D. 1.1)] in patients who were still on TNF-blocking agents (PP anti-TNF use) and even lower [2.7 (S.D. 1.1)] in patients who were still on their first

Discussion
In this study, we provided results from the DREAM biologics register, regarding the 5-year drug survival and 5-year outcomes on disease activity, daily functioning, quality of life and serious adverse events of TNF-blocking agents. The data show a moderate to high (60%) drug

www.rheumatology.oxfordjournals.org

199

Wietske Kievit et al.

FIG. 2 Effectiveness of TNF-blocking agents considering disease activity, functional disability and quality of life in 5 years of follow-up.

5.5 5.0 4.5

Disease activity
Intention to treat Per protocol first anti-TNF Per protocol anti-TNF use

1.4 1.3 1.2

Functional disability
Intention to treat Per protocol first anti-TNF Per protocol anti-TNF use

DAS-28

4.0 3.5 3.0 2.5 2.0 0 3 6 9 12 18 24 30 36 42 48 54 60

HAQ-DI

1.1 1.0 0.9 0.8 0.7 0 3 6 9 12 18 24 30 36 42 48 54 60

C 0.80
0.75

Time, months

Generic quality of life

Time, months
42 40 38

Downloaded from http://rheumatology.oxfordjournals.org/ by guest on September 21, 2012

Physical component scale

EQ5D utility score

SF-36 PCS
Intention to treat Per protocol first anti-TNF Per protocol anti-TNF use

0.70

36 34 32 30 28 Intention to treat Per protocol first anti-TNF Per protocol anti-TNF use

0.65

0.60

0.55

0.50 0 3 6 9 12 18 24 30 36 42 48 54 60

0 3 6 9 12

18

24

30

36

42

48

54

60

Time, months

Time, months

FIG. 3 Time till first serious infection or first malignancy during 5 years of follow-up.
1.00

Adverse events

Probability on first event

0.98

0.96

0.94

0.92 0.90 0.00 0

Serious infections Malignancies

Time, years

survival accompanied by good long-term effects of TNF-blocking agents in routine care. Patients had sustained low disease activity, normalized function and quality of life ratings reaching values similar to the general population. Furthermore, TNF-blocking agents appeared

relatively safe with a low incidence of serious infections and malignancies. To the best of our knowledge, this is the first article that presents 5-year data from routine care concerning not only a disease activity measure but also function and quality of life measures. There are three open extensions (ranging from 4 to 7 years) of original RCTs [2325]; two with etanercept-treated patients [23, 24]; and one with adalimumab-treated patients [25]. There are several European biologic registers reporting mainly on the occurrence of adverse events in patients treated with TNFblocking agents [13]. The DREAM register and the Danish biologic register (DANBIO) both collect detailed outcome data regarding RA disease activity besides the occurrence of adverse events, whereas the DREAM register also collects data on function and quality of life. The DANBIO register reported on long-term drug survival that ranged from 41 to 56% after 4 years of follow-up, depending on type of TNF-blocking agents [26], which seems similar to the 45% drug survival after 5-year follow-up reported in our study. However, the monoclonal antibody adalimumab (D2E7) in rheumatoid arthritis (ARMADA) open-label extension trial [25] reports a much longer drug survival of 64% after 4 years of follow-up. The latter can be due to stricter adherence to

200

www.rheumatology.oxfordjournals.org

Five-year follow-up on TNF blockers from the DREAM register

treatment in trials, which illustrates the need for reports from routine care. Long-term outcomes are only reported in the RCT extensions; mean DAS-28 scores ranged from 2.6 to 3.2 and mean HAQ scores ranged from 0.7 to 1.1 after at least 3 years of follow-up [2325]. These values seem to be similar to the mean DAS-28 score of 3.2 and a mean HAQ score of 0.96 after 5 years of follow-up in our population. The incidence rate for malignancies in our population (0.6/100 patient-years) is also comparable to rates reported by the open-label extensions [23, 24] and reports from other biologic registers [27, 28], showing rates ranging from 0.6 to 1/100 patient-years. The infection rate found in our data (2.9/100 patient-years) is lower than that reported in the British [29] and German [30] biologic registers and in the RCT extensions [2325] with rates ranging from 4.2 to 6.4/100 patient-years. An explanation of this difference in infection rates may lie in the self-reportage of adverse events in the DREAM register, which is probably an under reportage. We relied on the fact that recall for serious events defined as hospitalization would be good. Nonetheless, linkage with other national registers as is done in the Danish biologic and Swedish registers, would improve the study of serious adverse events. Besides the verification of RCT results in routine care, valuable information can be retrieved from biologic registers like DREAM as it includes a wide range of unselected patients and practice behaviour. A good example of an observation that can only be made by analysis of routine care data and not in a clinical trial is the extent to which advice originating from treatment guidelines is followed, as was analysed for the cessation of a treatment with a TNF-blocking agent in the case of a reported nonresponse after 3 months of treatment [31]. Other examples are: analysis of the effectiveness of TNFblocking agents in the patient groups that are mostly excluded from RCT, for example, the elderly [32, 33]; the effectiveness of switching between TNF-blocking agents [34, 35]; and the occurrence and effectiveness of dose escalation of TNF-blocking agents [36, 37]. Furthermore, the use of observational research data for the evaluation of the cost-effectiveness for reimbursement authorities is propagated [3840]. In The Netherlands, authorities are developing legislation for reimbursement decisions [41]. A provisional licence for reimbursement will be provided on the basis of favourable results from a pharmacoeconomic modelling study or a pharmaco-economic study alongside a RCT. Then, additional (pharmacoeconomic) data have to be collected in clinical practice over 3 years including drug survival, effectiveness rather than efficacy and costs based on true doses. A decision about a definitive licence for reimbursement will be based on those data. Because we have been expanding the register to all other biologic agents and to DMARD treatment (protocolized) in early RA patients, data from the DREAM register can be of great value in the context of reimbursement decisions. Besides the added value of register data there are also limitations. If one is interested in comparing treatments

regarding intended effects, the most important ones are: patients are not randomly assigned between treatment options and patients may be exposed to many different treatments in different orders that limit treatment comparisons. A comparison between different biologics in register data can introduce confounding by indication and thereby differences at baseline in important prognostic factors. For unintended effects (adverse events), however, confounding by indication is hardly a problem. It is suggested that it is possible in observational studies to correct for prognostic differences at baseline by either multiple regression analyses or a propensity score [42, 43]. However, by definition, it is not possible to correct for unmeasured confounding factors. This pleads for the use of RCT data, but then the problem of limited generalizability of RCT data is not solved. A solution could be a pragmatic randomized trial in which patients, not selected by stringent inclusion criteria, are randomly assigned at baseline [44] followed by a treatment that is not protocolized but at the discretion of the attending physician. However, the latter points to the other major limitation in routine care data. When patients do stop treatment with the first TNF-blocking agent, they can switch to any possible treatment option, which is in itself a valuable observation. When analysing long-term effectiveness, the mix of possible treatments at, for example, 5 years of follow-up makes the interpretation of results difficult. A solution can be to present both ITT and PP analyses, as was done in this article. Another limitation of using routine care databases for analysis is that you can only analyse data that are routinely collected and put in the database. In our case, we would have liked to have access to smoking status data of the patients because recent reports in literature showed that smoking can have a negative influence on the effect of therapy [45, 46]. Following on that, smoking status should be added to the DREAM data set. For a routine care database it is insurmountable that certain parameters of new interest are missing for analysis. The last limitation of routine care databases is that the number of patients who were initiated on certain therapies of interest is spread over years. Although 1560 patients were initiated on TNF-blocking therapy in the DREAM database, 174 patients had 5 years of follow-up at the time of analysis. The number 174 seems sufficient to describe means as in this article, but future analyses could give other results because of time trends in characteristics of the patients initiated on therapy. Lastly, we would like to point out that cost-effectiveness of TNF-blocking agents compared with the first best alternative was not considered in this article because of a lack of that control group treated with a first best alternative. Nonetheless, the literature shows varying costeffectiveness ratios in the range from very acceptable to unacceptable [47] and the acceptability of the costeffectiveness ratios varies with different assumptions. The ratios mostly result from modelling studies in which the long-term effects are extrapolated from the short-term effect because of the lack of long-term results at the time

Downloaded from http://rheumatology.oxfordjournals.org/ by guest on September 21, 2012

www.rheumatology.oxfordjournals.org

201

Wietske Kievit et al.

of analysis. In future research, those initial modelling studies could be updated, for instance, using long-term data from the DREAM registry on the effects of TNF-blocking agents on cost savings due to maintained work ability and the avoidance of expensive orthopaedic surgeries mediated by improved function. In conclusion, data from routine care registers have added value to data from RCTs like long-term effectiveness and safety data as reported in this article. We showed that TNF-blocking agents are used for a long time in individual patients and that this is accompanied by sustained low disease activity, functional normality and quality of life similar to that in the general population. Furthermore, it is shown that TNF-blocking agents are safe in routine care patients with low rates of serious infections and malignancies. Therefore, we conclude that the benefit to risk ratio for long-term TNF blocking therapy remains favourable. Rheumatology key messages Long-term TNF-blocking therapy is accompanied by good clinical outcomes and normalized quality of life. . The benefit to risk ratio for long-term TNF-blocking therapy remains highly favourable. . Data from routine care registers have added value to data from RCTs.
.

rheumatoid arthritis: results of a multicenter, randomized, double-blind, placebo-controlled trial. Arthritis Rheum 2004;50:35363. 4 Maini R St, Clair EW, Breedveld F et al. Infliximab (chimeric anti-tumour necrosis factor alpha monoclonal antibody) versus placebo in rheumatoid arthritis patients receiving concomitant methotrexate: a randomised phase III trial. Lancet 1999;354:19329. 5 Navarro-Sarabia F, Ariza-Ariza R, Hernandez-Cruz B, Villanueva I. Adalimumab for treating rheumatoid arthritis. J Rheumatol 2006;33:107581. 6 Weinblatt ME, Kremer JM, Bankhurst AD et al. A trial of etanercept, a recombinant tumor necrosis factor receptor: Fc fusion protein, in patients with rheumatoid arthritis receiving methotrexate. N Engl J Med 1999;340:2539. 7 Wolfe F, Michaud K, Dewitt EM. Why results of clinical trials and observational studies of antitumour necrosis factor (anti-TNF) therapy differ: methodological and interpretive issues. Ann Rheum Dis 2004;63(Suppl. 2):ii137. 8 Kievit W, Fransen J, Oerlemans AJ et al. The efficacy of anti-TNF in rheumatoid arthritis, a comparison between randomised controlled trials and clinical practice. Ann Rheum Dis 2007;66:14738. 9 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;54:3399407. 10 Miettinen OS. The need for randomization in the study of intended effects. Stat Med 1983;2:26771. 11 Brennan A, Bansback N. Reply, Wolfe et al. Do rheumatology cost-effectiveness analyses make sense?. Rheumatology 2004;43:6778. 12 Rothwell PM. External validity of randomised controlled trials: to whom do the results of this trial apply? Lancet 2005;365:8293. 13 Zink A, Askling J, Dixon WG, Klareskog L, Silman AJ, Symmons DP. European biologicals registers: methodology, selected results and perspectives. Ann Rheum Dis 2009;68:12406. 14 DREAM. www.dreamregistry.nl (29 September 2010, date last accessed). 15 Arnett FC, Edworthy SM, Bloch DA et al. The American Rheumatism Association 1987 Revised Criteria for the Classification of Rheumatoid Arthritis. Arthritis Rheum 1988;31:31524. 16 Fries JF, Spitz PW, Young DY. The dimensions of health outcomes - the health assessment questionnaire, disability and pain scales. J Rheumatol 1982;9:78993. 17 Zandbelt MM, Welsing PMJ, van Gestel AM, Van Riel PLCM. Health assessment questionnaire modifications: is standardisation needed? Ann Rheum Dis 2001;60: 8415. 18 Prevoo MLL, VantHof MA, Kuper HH, Vanleeuwen MA, VandePutte LBA, Vanriel PLCM. Modified disease-activity scores that include 28-joint counts - development and validation in a prospective longitudinal-study of patients with rheumatoid-arthritis. Arthritis Rheum 1995;38:448. 19 VanGestel AM, Prevoo MLL, VantHof MA, VanRijswijk MH, VandePutte LBA, Vanriel PLCM. Development and validation of the European League Against Rheumatism response criteria for rheumatoid arthritis - comparison with

Downloaded from http://rheumatology.oxfordjournals.org/ by guest on September 21, 2012

Acknowledgements
We are indebted to all research nurses and rheumatologist of the 13 departments of rheumatology for their participation and contribution to the data collection. Furthermore, we would like to acknowledge Thea van Gaalen, Lia Schalkwijk and Carien Versteegden for the data processing and Hans Groenewoud for the assistance in managing the database. We would also like to thank Dr Marjonne Creemers and Dr Annelies van Eden for training the research nurses in assessing joint counts Funding: Funding from the Dutch affiliations of Wyeth Pharmaceuticals, Abbott Pharmaceuticals, ScheringPlough Corporation, Roche Pharmaceuticals, UCB Pharma and Bristol-Myers Squibb enabled the data collection for the DREAM cohort. Disclosure statement: The authors have declared no conflicts of interest.

References
1 Blumenauer B, Judd M, Wells G et al. Infliximab for the treatment of rheumatoid arthritis. Cochrane Database Syst Rev 2002;3:CD003785. 2 Blumenauer B, Judd M, Cranney A et al. Etanercept for the treatment of rheumatoid arthritis. Cochrane Database Syst Rev 2003;4:CD004525. 3 Keystone EC, Schiff MH, Kremer JM et al. Once-weekly administration of 50 mg etanercept in patients with active

202

www.rheumatology.oxfordjournals.org

Five-year follow-up on TNF blockers from the DREAM register

the preliminary American College of Rheumatology and the World Health Organization International League Against Rheumatism criteria. Arthritis Rheum 1996;39: 3440. 20 EuroQol Group. EuroQola new facility for the measurement of health-related quality of life. The EuroQol Group. Health Policy 1990;16:199208. 21 Ware JE, Sherbourne CD. The Mos 36-item short-form health survey (SF-36).1. Conceptual-framework and item selection. Med Care 1992;30:47383. 22 Ware JE Jr, Kosinski M, Bayliss MS, McHorney CA, Rogers WH, Raczek A. Comparison of methods for the scoring and statistical analysis of SF-36 health profile and summary measures: summary of results from the Medical Outcomes Study. Med Care 1995;33(Suppl 4): AS26479. 23 Klareskog L, Gaubitz M, Rodriguez-Valverde V, Malaise M, Dougados M, Wajdula J. A long-term, open-label trial of the safety and efficacy of etanercept (Enbrel) in patients with rheumatoid arthritis not treated with other disease-modifying antirheumatic drugs. Ann Rheum Dis 2006;65:157884. 24 Moreland LW, Weinblatt ME, Keystone EC et al. Etanercept treatment in adults with established rheumatoid arthritis: 7 years of clinical experience. J Rheumatol 2006;33:85461. 25 Weinblatt ME, Keystone EC, Furst DE, Kavanaugh AF, Chartash EK, Segurado OG. Long term efficacy and safety of adalimumab plus methotrexate in patients with rheumatoid arthritis: ARMADA 4 year extended study. Ann Rheum Dis 2006;65:7539. 26 Hetland ML, Christensen IJ, Tarp U et al. Direct comparison of treatment responses, remission rates, and drug adherence in patients with rheumatoid arthritis treated with adalimumab, etanercept, or infliximab: results from eight years of surveillance of clinical practice in the nationwide Danish DANBIO registry. Arthritis Rheum 2010;62:2232. 27 Strangfeld A, Hierse F, Rau R et al. Risk of incident or recurrent malignancies among patients with rheumatoid arthritis exposed to biologic therapy in the German biologics register RABBIT. Arthritis Res Ther 2010;12:R5. 28 Askling J, van Vollenhoven RF, Granath F et al. Cancer risk in patients with rheumatoid arthritis treated with anti-tumor necrosis factor alpha therapies: does the risk change with the time since start of treatment? Arthritis Rheum 2009;60: 31809. 29 Dixon WG, Watson K, Lunt M, Hyrich KL, Silman AJ, Symmons DP. Rates of serious infection, including site-specific and bacterial intracellular infection, in rheumatoid arthritis patients receiving anti-tumor necrosis factor therapy: results from the British Society for Rheumatology Biologics Register. Arthritis Rheum 2006; 54:236876. 30 Listing J, Strangfeld A, Kary S et al. Infections in patients with rheumatoid arthritis treated with biologic agents. Arthritis Rheum 2005;52:340312. 31 Kievit W, Fransen J, Adang EM et al. Evaluating guidelines of continuation of anti-TNF treatment after three months: clinical effectiveness and costs of observed care and different alternative strategies. Ann Rheum Dis 2009;68: 8449.

32 Radovits BJ, Kievit W, Fransen J et al. Influence of age on the outcome of anti-TNF alpha therapy in rheumatoid arthritis. Ann Rheum Dis 2009;68:14703. 33 Genevay S, Finckh A, Ciurea A, Chamot AM, Kyburz D, Gabay C. Tolerance and effectiveness of anti-tumor necrosis factor alpha therapies in elderly patients with rheumatoid arthritis: a population-based cohort study. Arthritis Rheum 2007;57:67985. 34 Blom M, Kievit W, Fransen J et al. The reason for discontinuation of the first tumor necrosis factor (TNF) blocking agent does not influence the effect of a second TNF blocking agent in patients with rheumatoid arthritis. J Rheumatol 2009;36:21717. 35 van Vollenhoven RF, Harju A, Brannemark S, Klareskog L. Treatment with infliximab (Remicade) when etanercept (Enbrel) has failed or vice versa: data from the STURE registry showing that switching tumour necrosis factor alpha blockers can make sense. Ann Rheum Dis 2003;62: 11958. 36 Blom M, Kievit W, Kuper HH et al. The frequency and effectiveness of dose increase of adalimumab, etanercept and infliximab in daily clinical practice. Arthritis Care Res 2010;62:133541. 37 Zink A, Listing J, Strangfeld A, Gromnica-Ihle E, Demary W, Schneider M. Dose adjustment in patients treated with infliximab in routine rheumatologic care in Germany. Results from the Biologics Register RABBIT. Z Rheumatol 2006;65:4416. 38 Wolfe F, Michaud K, Pincus T. Do rheumatology cost-effectiveness analyses make sense? Rheumatology 2004;43:46. 39 Maetzel A. Cost-effectiveness analysis: out of touch with clinical reality? Arthritis Rheum 2005;53:34. 40 Farahani P, Levine M, Goeree R. A comparison between integrating clinical practice setting and randomized controlled trial setting into economic evaluation models of therapeutics. J Eval Clin Pract 2006;12:46370. 41 Pronk MH, Lelie van der LG, Schoonderbeek JRA. Dure pillen anders vergoeden. Medisch Contact 2007;62: 17681770. 42 Klungel OH, Martens EP, Psaty BM et al. Methods to assess intended effects of drug treatment in observational studies are reviewed. J Clin Epidemiol 2004; 57:122331. 43 Rosenbaum PR, Rubin DBj. The central role of the propensity score in observational studies for causal effects. Biometrika 1983;70:4155. 44 Schwartz D, Lellouch J. Explanatory and pragmatic attitudes in therapeutical trials. J Chronic Dis 1967;20: 63748. 45 Abhishek A, Butt S, Gadsby K, Zhang W, Deighton CM. Anti-TNF-alpha agents are less effective for the treatment of rheumatoid arthritis in current smokers. J Clin Rheumatol 2010;16:158. 46 Mattey DL, Brownfield A, Dawes PT. Relationship between pack-year history of smoking and response to tumor necrosis factor antagonists in patients with rheumatoid arthritis. J Rheumatol 2009;36:11807. 47 Welsing PM, Kievit W, Laan RF, Severens JL. Quality of life and costs for different treatment strategies for rheumatoid arthritis. Expert Rev Pharmacoecon Outcomes Res 2005; 5:395410.

Downloaded from http://rheumatology.oxfordjournals.org/ by guest on September 21, 2012

www.rheumatology.oxfordjournals.org

203