Welwyn Garden City
Tel: 01707 363636
Fax: 01707 363690
1 February, 2006
71 High Holborn
RE: WEST MIDLANDS HEALTH TECHNOLOGY ASSESSMENT COLLABORATION (WMHTAC)
ASSESSMENT REPORT ON ADALIMUMAB, ETANERCEPT AND INFLIXIMAB FOR THE TREATMENT OF
RHEUMATOID ARTHRITIS (RA).
Schering-Plough welcomes the opportunity to comment on this report and its technical content.
Following a thorough review of the WMHTAC report, this letter sets out our comments – firstly a
summary of what we perceive to be the critical issues, and subsequently more detailed information
relating to these and other related issues.
Therefore, we would first like to draw attention to a number of key issues of concern that we
believe are critical:
• Costs: The estimation of treatment costs associated with infliximab in the WMHTAC model
is flawed. Firstly, the drug acquisition cost for infliximab used in the model as stated in the
HTA report, is incorrect. The correct NHS price for infliximab is £419.62 per 100 mg vial.
If the correct NHS price is used, for a typical patient requiring 3 vials or less per infusion,
the treatment costs associated with infliximab are lower than the comparable costs for other
anti-TNF treatments after the induction dosing period, even assuming vial wastage.
Secondly, in clinical practice, where such vial wastage is avoided, the difference in
treatment costs between infliximab and other anti-TNF treatments will be even greater. We
would ask for confirmation regarding the price used in the model and emphasise that the
ICERs estimated by the WMHTAC model are highly unreliable.
• Efficacy: Infliximab is at least as efficacious as other anti-TNFs. While this may be
obscured by enrollment differences in severity of underlying disease (limiting ability to
show benefit) between the various trials, the case is clear that equal efficacy has been
achieved with a greater patient severity in the infliximab trials. Since no evidence has been
generated since NICE’s 2002 guidance to support the clinical superiority of any biologic in
RA we argue that the conclusions of the WMHTAC HTA report appear perverse.
• Safety: Infliximab has equal safety compared to other anti-TNFs. Recent data from the
BSRBR show that there are no significant differences in the estimated incidence rates of
serious infections between infliximab, etanercept, and adalimumab.
• Treatment Sequence vs. Other DMARDS: The WMHTAC HTA presents the results of a
sequential analysis including DMARDs and anti-TNF treatments. Firstly, we would argue
that this approach represents a significant departure from the agreed scope of this appraisal,
thus rendering its conclusions perverse. In addition, we would argue that this sequential
approach is invalid owing to the absence of relevant clinical data to enable reliable
economic modeling. While the HTA acknowledges that the assumptions underpinning this
sequential analysis are a “limitation”, we feel this characterization trivializes the true nature
of the limitation that, in reality, invalidates the WMHTAC analyses.
Detailed explanation regarding key concerns:
We will address each of these points in some detail below, along with other comments on
clinical issues with the HTA report, the critique of Schering-Plough’s original submission, and
further comments on the WMHTAC economic evaluation.
1. Incorrect price for infliximab used: In Table 42, page 197, it should be noted that the cost
of infliximab used in the WMHTAC model is incorrect. Table 42 lists the price as £451.20
per 100 mg vial, but the correct cost is £419.62 per 100 mg vial (British Medical
Association 2005). The WMHTAC overestimates the cost of infliximab (not only based on
price, but also see point 2 below) and this will have an impact on any economic evaluation.
This issue is critical in relation to the conclusions drawn by the WMHTAC team and we
would argue that the ICERs estimated by the WMHTAC model are highly unreliable, due in
part to this error.
2. Dosage and vial optimisation: In Table 42, page 197, the infliximab cost is estimated based
on a 70 kg patient, with drug wastage, when a full vial is not used. This requires all patients
to receive 3 infliximab vials. As stated previously in the Schering-Plough submission, the
infliximab start-up regimen requires additional doses within the first 12 weeks as
recommended by the BSR. However, upon entering maintenance after the first 3 months,
the cost advantage is clearly apparent for the patients needing 3 or fewer vials per infusion
of infliximab. In any situation with 3 or fewer vials being used, infliximab will be less
expensive than the other anti-TNF inhibitors after the induction dosing period, even
including infusion costs (Table 1).
Reference: Schering-Plough Page 2 of 12
If approved RA dosing (3 mg/kg) for infliximab is used, patients up to 100 kg would use 3
vials per infliximab infusion. In the British Society for Rheumatology Biologics Registry
(BSRBR) report submitted to NICE with the Schering-Plough infliximab RA submission
(Schering-Plough, 2005), 42% of patients exposed to infliximab weighed below 67 kg, and
94% were below 100 kg. Thus, the majority of patients in the BSRBR could potentially
receive the less expensive infliximab maintenance treatment when compared to treatment
with the other biologics. Furthermore, the 42% of patients weighing less than 67 kg would
use only 2 infliximab vials with label dose, providing an even less expensive proposition for
infliximab compared to the other biologics (Table 2).
It is important to note that many patients for whom infliximab proves to be less expensive,
are easily identified prior to treatment by their weights. Thus, our analysis does not rely on
an artificial construct such as a “responders only” analysis, but rather on an easily
determined demographic. Additionally, in clinical practice, where vial wastage is avoided,
the cost advantage favouring infliximab over the other anti-TNF treatments will be even
greater. Sharing vials in this way could reduce the cost of infliximab, thereby improving its
cost-effectiveness. However, no reference is made in the HTA (and no sensitivity analyses
were run) with respect to this vial optimisation, which is best practice (Norfolk and
Norwich,University Hospital, unpublished, personal communication. Submitted in
confidence). This further highlights the questionable nature of the ICERs produced by the
WMHTAC model. In the context of vial optimization, the illustrative cost estimates in
Tables 1 and 2 below are conservative, since these calculations assume drug wastage.
Reference: Schering-Plough Page 3 of 12
Table 1. Patients Needing 3 Vials of Infliximab per Infusion (assuming drug wastage)
Initial trial period (3 months) Annual cost (after initial 3 months) Total costs
Treatment – Acquisition Administration Monitoring Acquisition Administration Monitoring Initial Subsequent
Dosage drug cost cost costs drug cost cost costs trial annual costs
Adalimumab 2,145.60a 246.00b 127.91c 9,297.60a 0.00 205.08c 2,519.51 9,502.68
Etanercept 2,145.12d 246.00e 127.91f 9,295.52 0.00 205.08f 2,519.03 9,500.60
3mg/kg, 3 3,776.58g 372.00h 127.91f 8,182.59 806.00h 13.08f 4,276.49 9,001.67
Table 2. Patients Needing 2 Vials of Infliximab per Infusion (assuming drug wastage)
Initial trial period (3 months) Annual cost (after initial 3 months) Total costs
Treatment – Acquisition Administration Monitoring Acquisition Administration Monitoring Initial Subsequent
Dosage drug cost cost costs drug cost cost costs trial annual costs
Adalimumab 2,145.60a 246.00b 127.91c 9,297.60a 0.00 205.08c 2,519.51 9,502.68
Etanercept 2,145.12d 246.00e 127.91f 9,295.52d 0.00 205.08f 2,519.03 9,500.60
3mg/kg, 3 2,517.72g 372.00h 127.91f 5,455.06g 806.00h 13.08f 3,017.63 6,274.14
a. Medication costs for adalimumab estimated at £357.60 per dose based on WMHTAC report; during maintenance period, assume
patients receive 26 adalimumab doses per annum.
b. Administration costs estimated from CRD/CHE Technology Assessment Group psoriatic arthritis HTA, Section 10.12.
c. Monitoring costs estimated from CRD/CHE Technology Assessment Group psoriatic arthritis HTA, Section 10.12.
d. Medication costs for etanercept estimated at £89.38 per dose based on WMHTAC report; during maintenance period, assume patients
receive 104 etanercept doses per annum.
e. Administration costs taken from CRD/CHE Technology Assessment Group psoriatic arthritis HTA, Section 10.12.
f. Monitoring costs taken from CRD/CHE Technology Assessment Group psoriatic arthritis HTA, Section 10.12.
g. Medication costs for infliximab estimated at £897.99 per infusion based on £419.62 per 100 mg vial (British Medical Association 2005);
during maintenance period, patients receive 6.5 infliximab infusions per annum.
h. Administration cost of £124 per infusion based on previous NICE Health Technology Assessments (Brennan et al. 2005, Barton et. al.
1. No superiority evidence for any anti-TNF is available: As no significant new data on
anti-TNFs vs. traditional DMARDS has been introduced since the last NICE assessment, the
previous NICE guidance in “late” RA patients should hold in terms of its general placement
of anti-TNFs in a DMARD sequence (after at least 2 DMARD failures- NICE Guidance,
March 2002). No evidence has been generated since the 2002 guidance to support the
clinical superiority of one biologic therapy over another. Clinical experience, a meta-
analysis by Hochberg et al (Ann Rheum Dis 2003) and data from the BSRBR have all
suggested similar efficacy amongst TNF antagonists. While this may be obscured by
enrollment differences in severity of underlying disease (limiting ability to show benefit)
between the various trials, the case is clear that equal efficacy has been achieved with a
greater patient severity in the infliximab trials. It is clear that greater severity is a marker for
less ability to show improvement. Equal efficacy with such bias in enrollments, implies
infliximab is at least the equal to the other anti-TNFs.
Reference: Schering-Plough Page 4 of 12
The summary statement of the WMHTAC report, suggesting that etanercept is the most cost
effective anti-TNF, lacks face validity and would baffle most rheumatologists.
As stated in the Schering-Plough infliximab RA NICE submission, it should be noted that
since the last NICE appraisal, studies have shown excellent clinical results from initiation of
infliximab therapy at earlier stages of RA, resulting in expanded labelling to include patients
with severe, active and progressive disease not previously treated with methotrexate or other
disease-modifying antirheumatic agents. When radiographic scores from baseline and 2
years were compared in the early RA subset of the ATTRACT trial, patients treated with
infliximab plus methotrexate have less progression in joint erosion than those treated with
methotrexate alone (Breedveld FC et al, Ann Rheum Dis 2004).
2. NICE’s previous 2002 guidance: In the previous review of infliximab in 2002, the
WMHTAC assessment team produced an ICER of £99,373 per QALY for infliximab, while
NICE ultimately estimated an ICER of the biologic therapies for RA to be in the region of
£27,000 to £35,000 per QALY, in a separate analysis for their final guidance (NICE, March
2002). Given that no new clinical superiority evidence has been generated, and since NICE
previously produced guidance showing that the biologic therapies (including infliximab) are
cost effective, it seems perverse that the current WMHTAC model has produced high ICER
results for infliximab, comparable to their previous estimates in 2002, and that etanercept is
favoured on cost-effectiveness grounds.
3. Importance of patient preference: In section 8, page 228, the HTA states “… the most
appropriate choice of anti-TNF inhibitor may depend on patient preference as to route of
administration.” We must reiterate that infliximab should be made available for patients
preferring infusion over self-injection. As stated on page 22 of the Schering-Plough
submission, a survey conducted by the UK National Rheumatoid Arthritis Society in
December 2002 (NRAS 2003) showed that 76% of respondents (in over 500 patients)
preferred intravenous infusion over subcutaneous injections, primarily because of the longer
time between infusions (2 months), and the presence of trained clinical personnel
administering it. Also from the Schering-Plough submission (page 22), a retrospective
Canadian analysis of the administrative claims data from the Saskatchewan Drug Plan and
Extended Benefits Branch database from September 2001 to September 2003 (Olszynski
WP et al, EULAR 2004) showed that only 24% of etanercept patients were 100% compliant
compared to 80% of infliximab patients (p <0.01). The percentage of patients receiving a
minimum of 80% of the number of prescriptions was also significantly higher (p <0.01) for
infliximab patients (91%) than for etanercept patients (71%). The association between
patient treatment satisfaction and long-term compliance is now well established. We would
argue that this represents a crucial consideration for patients expressing a preference for
intravenous infusion rather than self-injection, with clear implications for long-term disease
Reference: Schering-Plough Page 5 of 12
1. Similar safety profiles exist for biologic therapies: In section 3, page 41 of the HTA, it
states “The combination involving infliximab, however, was associated with an increased
risk of serious infection.” It should be noted that the BSRBR data indicates that serious
infections across all three biologics are similar in number. In a recent assessment of the
BSRBR, data on incidence of serious infection were analysed on 2602, 2871 and 915
consecutive patients treated respectively with etanercept, infliximab, and adalimumab
registered with the BSRBR for any rheumatic disease. The estimated incidence rate of all
serious infections on biologic agents was 50 - 65 cases per 1,000 patient-years, with no
significant difference between the anti-TNF agents (Dixon W et al, Rheumatology 2005).
Treatment Sequence of Infiximab vs. Other DMARDs:
1. Place of infliximab in relation to DMARD use in model is inconsistent with past NICE
Guidance and BSR Guidelines: In the WMHTAC model, it appears that infliximab and the
other anti-TNFs could have been used in variable places within a sequence of many
DMARDs- infliximab could have been given after a sequence of 8 DMARDs, for example.
Such a treatment approach is not consistent with both current British Society for
Rheumatology (BSR) guidelines for prescribing anti-TNFs in RA (Ledingham J et al,
Rheumatology 2005) and previous NICE guidance on the use of etancercept and infliximab
for the treatment of RA (NICE, March 2002). Both guidelines indicate that biologics
(specifically infliximab and etanercept for the NICE 2002 guidance) can be prescribed in
patients with clinically active RA that has not responded to at least 2 DMARDs, including
methotrexate (unless contraindicated). While “not responding to at least 2 DMARDs” could
be an option generated in the WMHTAC model, the fact that the model randomly positions
infliximab in various places within a sequence of potentially multiple DMARDs is not
aligned with real world clinical practice.
2. Biases against anti-TNFs: As stated above, WMHTAC’s analytic approach is to compare
sequences of DMARDs with the insertion of an anti-TNF at various points in the sequence.
The assessment team acknowledges the limitation of this approach by stating the “poor
quality of the data on effectiveness of conventional DMARDs” (page 215), yet this forms
the basis of comparison for the new and old drugs (section 4.2.2). We feel that this
limitation is so serious as to be more than just a limitation- it likely invalidates the model.
The approach severely biases the analysis against anti-TNFs in two ways.
First, this assumes that the clinical trial populations in the literature for DMARDs are
identical to those in the anti-TNF trials. On the contrary, examining the clinical
characteristics of patients suggests that those enrolled in contemporary trials are among the
top 25% of RA with the most severe disease based on nomograms of disease activity and
severity [Wolfe F, Am J Med, 1997]. This is further confirmed by data suggesting faster
HAQ progression in contemporary trials despite therapeutic improvements over the past 2
decades [Aletaha D, Ward M, Ann Rheum Dis in press]. Similarly, when looking at an early
RA cohort, patients with more than 6-10 swollen joints have a higher probability of
radiographic progression ranging from >82% to >93% depending on the presence or
absence of RF positivity or erosive joint damage (Drossaers-Bakker KW et al, Arth Rheum
2002). Therefore, the analysis is biased by comparing anti-TNF treatment to DMARD
treatment in a far less severe population.
Reference: Schering-Plough Page 6 of 12
Secondly, the WMHTAC analysis does not account for the lower likelihood of response for
anti-TNFs associated with having failed more prior DMARDS prior to their use. Thus,
regardless of when a DMARD is given in the sequence of DMARDs in the WMHTAC
model, literature-based response, efficacy and duration of therapy are identical for a given
DMARD. In the model, efficacy evidence for any DMARD from, say, first line treatment is
assumed to be also relevant for eighth line treatment (and anything in between). This is
despite the increased severity of disease with progression and the passing of time and lesser
ability of a conventional DMARD to influence the more severe disease. In such a model, it
is not clear what better assumption could be made, rather just that such an assumption is
erroneous. Such is the advantage of the Schering-Plough model, in that it is based on trial
data of actual observations of a particular treatment sequence.
Comparing the intervention and control arms directly from a randomized controlled trial
avoids the confounding that can occur from cross randomized trial comparisons and from
differences in the clinical population as was done in the Schering-Plough submission. In the
current WMHTAC model, the comparator for the anti-TNFs are from literature-based
DMARD treatments in clinical populations from earlier eras as opposed to the control arms
of the randomized trials.
In a recent review of varying analytic approaches for economic modeling of infliximab in
RA (including a review of 2 similar models to this WMHTAC analysis, also produced by
the Birmingham group), the authors indicate that 2 of the 4 reviewed models (Manufacturer
and Kobelt models) produced similar ICERs (£25,000 - £30,000/QALY). However, the 2
Birmingham models produced much higher ICERs, ranging from (£50,000 -
£60,000/QALY, to more than £100,000/QALY). The authors concluded that while the
Manufacturer and Kobelt models assumed that infliximab was given to patients who had
failed a mean of 2.5-2.8 DMARDs, in combination with methotrexate the differing sequence
of treatments for DMARDs and infliximab in the Birmingham models could potentially
account for the differences in cost-effectiveness produced between all 4 reviewed models
(Drummond et al, Med Decis Making 2005).
Clinical Issues with the HTA Assessment:
1. Non-differentiation of RA Clinical Subgroups: The assessment appears to confuse the
distinction among refractory, resistant and placebo treated RA (Kroot EJ et al, Baillieres
Best Pract Res Clin Rheumatol 1999). It treats methotrexate in the infliximab trials as being
identical to placebo, yet clinical data suggests that if methotrexate were withdrawn, disease
activity would dramatically flare (Kremer JM et al, Am J Med 1987). The application of the
trial data to the model in this manner may produce flawed results.
2. Trial heterogeneity: Related to point 2 in the section above, the assessment does not
distinguish among trials on the basis of disease duration or number of previous DMARD
failures, even though it recognizes that “as successive DMARDs are tried to control disease,
the likelihood of sustained drug use declines” (section 2.1.9, page 30). The various trials
were clearly differentiated by severity, but this was not accounted for in the HTA. The
comparisons have no validity.
Reference: Schering-Plough Page 7 of 12
Again, RA patients, who have had the disease for longer periods of time or equivalently who
have failed more prior DMARDS, are more difficult to treat. They have lower likelihood of
ACR20 response (35% for >10 years and 38% for 5-10 years), lower reductions in acute
phase reactants, shorter therapy retention rates, and lower HAQ effect sizes and response
rates [Aletaha D et al, Rheumatology 2002; Anderson JJ et al, Arthritis & Rheumatism 2000;
Aletaha D, Ward M, Ann Rheum Dis in press]. This is an admitted limitation of the
WMHTAC analysis (page 215). This inadequacy, however, confounds cross trial
comparisons because the clinical trial populations differ in the duration of disease, with
ATTRACT, for example, having the longest. This would imply that the patients in this trial
were the most difficult to treat and equating the trial populations, as done in the assessment,
does not acknowledge this. No attempt was made to adjust for these potential confounding
factors. Again, we feel that the problem is quite serious, to the point that the acknowledged
“limitation” may in fact wholly invalidate the WMHTAC model and approach.
Critique of the Schering-Plough economic evaluation:
1. Description of ARAMIS: The first criticism of the Schering-Plough submission surrounds
the use of patient populations seen in “private practices in the US and Canada between 1981
and 1995” and their relevance to practice in the NHS in 2005 (page 167). First, the patient
populations consist not only of those in private practices, but also regional ones (Santa Clara
and Saskatoon) as well as institutional ones such as Stanford. Second, while the health
economics of care differ substantially between the UK and North America, the data used in
the model are limited to the likelihood of progression of HAQ with commonly used
medications and change in medication, which should be much more likely to be similar
between North America and the UK. Third, ideally one would like to know the prognosis
for UK patients with RA in 2005, but follow up data will be unavailable for another 10 to 20
years. Hence, observational data from earlier eras (10-20 years earlier) by necessity are
needed to inform longitudinal projections. In particular, as noted in the assessment,
“Observational data are a more realistic representation of outcomes in practice and therefore
more suitable for CE analyses” (page 150), which is what ARAMIS provides.
2. Treatment switching: The report questions why patients in the comparator arm of the
model cannot switch to infliximab plus methotrexate. However, in the absence of such a
crossover being explicitly included in the trial, the likelihood of response would be
speculative and wrought with uncertainty with little scientific basis, much as with the
WMHTAC model, as we argue. Hence, the model does not consider that possibility without
the existence of any adequate data.
3. HAQ benefit with increasing age: The report questions the validity of the assumption that
patients enjoying sufficient benefit from infliximab to continue treatment would continue to
experience the same HAQ score as the age of the cohort increases. However, the worsening
in HAQ with age is likely to occur equally in all patients, regardless of treatment (as was
assumed within the WMHTAC model): this means that the assumption of constant HAQ
would have no impact on the results, since any age-related worsening would affect the
infliximab arm and the comparator arm equally.
Reference: Schering-Plough Page 8 of 12
4. Double counting of HAQ improvements: The report asserts that HAQ improvements are
double counted by taking account of both radiographic progression and direct improvements
on HAQ. As to the modeling of radiographic benefit, smallest detectable difference is
indeed a statistical concept, but it is endorsed by Outcome Measures in Rheumatology
(OMERACT). It provides a starting point to assess reliably the potential clinical impact of
such progression. The use of less stringent statistical criteria would introduce potential error
in classification due solely to random variation by different readers. While it is possible that
there may be some double counting of HAQ benefits due to radiographic stabilization, the
analysis is done so that the absolute benefit in HAQ (0.27) is identical to that observed in the
observational cohort of patients with early RA. This benefit then is not in addition to
observed progression rates from the trial but already incorporates those HAQ progression
benefits to yield a net 0.27 HAQ benefit at 5 years.
Whether an 0.27 HAQ benefit is possible in patients with more long-standing disease is
unknown since no similar study has been done in non-early patients (page 168). The
assessment group’s implication that it would not be possible suggests that patients with
longer standing disease have a lower likelihood of HAQ benefit. If so, this decreased
likelihood of response should also be considered in cross trial comparison of response rates,
in particular for ATTRACT.
5. Application of radiographic change: The report questions whether the radiographic
changes seen in early RA also apply to the ATTRACT population or to patients who receive
a short course of treatment and show an ACR20 response at 14 weeks. However, the van
der Heijde scores of radiographic change measured during the ATTRACT and ASPIRE
studies are, in fact, similar for patients receiving infliximab. In ATTRACT, the median
change among patients receiving 3 mg/kg infliximab was 0.5 points (versus 4.0 within the
methotrexate group, p < 0.001), while the median change within the ASPIRE trial was 0.00
points among patients receiving 3 mg/kg infliximab and 0.43 among patients receiving
To clarify, any patients who were not ACR20 responders at week 14 had infliximab therapy
discontinued and did not receive any radiographic benefit. Patients who were ACR20
responders at week 14 would continue to receive infliximab in addition to the doses at
weeks 0, 2 and 6 and thus treatment would NOT be limited to 6 weeks for radiographic
benefit to occur.
6. Criticisms of VAS: Although the VAS utility from ATTRACT and ARAMIS in the Wong
publication is criticized for not being truly preference based (page 152), the use of EQ-5D
would have improved (lowered) the cost-effectiveness ratios. Thus, despite the comment
that those “results should be treated with caution”, the results would have been even more
favorable to infliximab with EQ-5D. One advantage of this approach is the use of actual
patient expressed values from the trial as opposed to external estimates for patients with
given HAQ categories.
7. Infliximab dosing: The report suggested that combining the results of different doses of
infliximab may bias the analysis in favour of treatment since patients receiving 10 mg/kg
infliximab responded better than those receiving other doses. However, post hoc analyses
based on the ATTRACT study (Lipsky et al 2000) demonstrate that there was no significant
difference in the proportion of patients achieving ACR 70, ACR 50 or ACR 20 between
patients receiving 3 mg/kg infliximab every eight weeks and those receiving 10 mg/kg
infliximab every eight weeks (p > 0.05).
Reference: Schering-Plough Page 9 of 12
8. Criticism of stopping rule: The report suggested that the Schering-Plough submission
inappropriately used a stopping rule based on ACR 20, rather than DAS28 as recommended
by NICE. Although DAS28 is endorsed by NICE, there is, as yet, very little data from
clinical trials demonstrating the proportion of patients who achieve this clinical endpoint,
since this measure has only recently been developed. Indeed, the WMHTAC group also
failed to use the DAS28 measure and, instead, used the proportion of patients discontinuing
early due to "lack of efficacy" from a variety of trials, which are likely to have used very
different definitions varying from one another and from the criteria likely to be used in UK
Comments on the WMHTAC economic evaluation:
1. Assumptions on HAQ improvement may be flawed: The WMHTAC model assumes that
the HAQ improvement experienced through treatment is proportional to baseline HAQ
score. The HAQ scale is based on a questionnaire in which participants rank their disease
severity in a number of domains. Although the average score and standard deviation are
frequently presented, HAQ scores are in fact ordinal rather than numeric, which means that
any calculations based on the mean and standard deviation or the percentage change in HAQ
are inappropriate. It is therefore inappropriate to apply a multiplicative function whereby
the improvement in HAQ is assumed to be proportional to baseline HAQ. The report also
provides no justification supporting this assumption or justifying why the benefit from
treatment would not more accurately be modelled by calculating the absolute change in
HAQ score regardless of baseline disease severity, or by assuming a different relationship
between baseline scores and the level of improvement achieved through treatment. If the
true relationship between baseline disease severity and HAQ differed from the relationship
assumed within this analysis, it is likely that some systematic bias could be introduced,
particularly since the patient population included within the infliximab trials may have had
more severe disease and a wider variety of disease severity levels than the population
participating in trials on etanercept and adalimumab.
2. Calculation of HAQ scores may be flawed: The calculation of the parameters defining the
beta distribution (a and b) appears to have been conducted using the mean and standard
deviation around the baseline HAQ and the absolute improvement in HAQ, which again are
inappropriate parameters for ordinal data such as this. WMHTAC also assumed that these
ordinal values followed a normal distribution, despite the need to adjust scores to create
plausible values and the large number of patients with zero HAQ scores at baseline
(indicating that their "moderate-severe RA" caused no disruption to their daily lives). The
use of ordinal data in this way and the assumption of normality may have introduced bias
into the calculation of HAQ scores and therefore into the calculation of utility gains through
3. Discontinuation of treatment: The proportion of patients discontinuing treatment due to
side-effects or lack of efficacy is based on the results of various different clinical trials,
which are likely to have used different criteria for determining whether patients would cease
treatment due to lack of efficacy or side-effects. The nature of the analysis means that these
differences between protocols are assumed to reflect differences between treatments in
routine clinical practice.
Reference: Schering-Plough Page 10 of 12
4. Monitoring assumptions are unclear: Within Table 43, page 198, infliximab-treated
patients are assumed to receive a full blood count, ESR and biochemical profile at every
infusion (every eight weeks once on a stable dose), whereas patients receiving etanercept or
adalimumab receive these tests every three months. If this assumption is not justified, it
would bias the analysis against infliximab. In addition, the three-monthly checkups for
patients receiving etanercept and adalimumab should require an additional consultation with
a nurse or clinician, whereas monitoring infliximab patients would require minimal nurse-
time to take a blood sample from a patient is already within the clinic receiving the infusion.
It is unclear from the report whether these additional consultations for patients receiving
etanercept or adalimumab are included within the reported costs. Similarly, the report does
not state how often patients receiving etanercept or adalimumab would require a visit by a
5. Anti-TNF benefits are not captured completely: While the WMHTAC model facilitates
the examination of a sequence of drugs, the underlying data regarding therapeutic response
and the comparator to the anti-TNFs have numerous flaws as noted above. In addition, the
only costs considered are the cost of therapy (page 199) and thus the WMHTAC model does
not consider any economic offsets from the prevention of disability. The natural history of
disease model focuses on duration of therapy and time to withdrawal. HAQ progresses
inexorably, so all patients do so regardless of therapy, and in fact, HAQ progresses even
during therapy (page 195). Thus, there is no chance for any patients to have a prolonged
remission. The only benefit modeled is a temporary incremental improvement in HAQ
during therapy but that degrades over time because HAQ nonetheless progresses even with
therapy. On discontinuing treatment, patients not only return to their baseline HAQ
(baseline prior to last therapy), but also deteriorate immediately to a HAQ level indicative of
no treatment (page 200).
Once again, we are grateful for the opportunity to comment on the WMHTAC assessment report,
and we look forward to continued dialogue with NICE regarding the issues raised in this letter.
Director, Communications & Public Affairs
Reference: Schering-Plough Page 11 of 12
1) Aletaha D, Smolen DS. The rheumatoid arthritis patient in the clinic: comparing more than 1,300 consecutive
DMARD courses. Rheumatology 2002; 41(12): 1367-74.
2) Aletaha D, Ward M. Duration of rheumatoid arthritis influences the degree of functional improvement in clinical
trials. Ann Rheum Dis 2005; [Epub ahead of print].
3) Anderson JJ, Wells G, Verhoeven AC, Felson DT. Factors predicting response to treatment in rheumatoid
arthritis: the importance of disease duration. Arthritis Rheum 2000; 43(1): 22-9.
4) Breedveld FC, Emery P, Keystone E, Patel K, Furst DE, Kalden JR, St Clair EW, Weisman M, Smolen J,
Lipsky PE, Maini RN. Infliximab in active early rheumatoid arthritis. Ann Rheum Dis 2004; 63 (2):149–155.
5) Brennan A, Bansback N, Nixon R. Modelling the cost-effectiveness of TNF-α inhibitors in the management of
rheumatoid arthritis: results from the British Society for Rheumatology Biologics Registry, Draft 1.0, 27 April
6) British Medical Association (2005). British National Formulary (BNF), No. 49. British Medical Association and
the Royal Pharmaceutoical Society of Great Britain, London
7) Dixon W, , K. Hyrich, K. Watson, A. Silman, D. Symmons, et al. Serious infection rates in patients receiving
biologic therapy in the United Kingdom: Results from the BSR Biologics Register (BSRBR). Rheumatology
2005; 44 (Supplement 1): i11, Abstract OP25.
8) Drossaers-Bakker KW, Zwinderman AH, Vlieland TP, Van Zeben D. Long-term outcome in rheumatoid
arthritis: a simple algorithm of baseline parameters can predict radiographic damage, disability, and disease
course at 12-year followup. Arthritis Rheum 2002; 47 (4): 383-90.
9) Drummond MF, Barbieri M, Wong JB. Analytic choices in economic models of treatments for rheumatoid
arthritis: What makes a difference? Med Decis Making 2005; 25: 520-533.
10) Hochberg MC, Tracy JK, Hawkins-Holt M, Flores RH. Comparison of the efficacy of the tumour necrosis factor
alpha blocking agents adalimumab, etanercept, and infliximab when added to methotrexate in patients with
active rheumatoid arthritis. Ann Rheum Dis 2003; ;62 Suppl 2:ii13-6.
11) Kremer JM, Rynes RI, Bartholomew LE.. Severe flare of rheumatoid arthritis after discontinuation of long-term
methotrexate therapy. Double-blind study. Am J Med 1987; 82 (4): 781-6.
12) Kroot EJ, van de Putte LB, van Riel PL. Management of therapy-resistant rheumatoid arthritis. Baillieres Best
Pract Res Clin Rheumatol 1999;13 (4): 737-52.
13) Ledingham J, Deighton C. Update on the British Society for Rheumatology guidelines for prescribing TNFα
blockers in adults with rheumatoid arthritis (update of previous guidelines of April 2001). Rheumatology 2005;
14) Lipsky PE, van der Hejde DM, St. Clair EW, Furst DE, et al. Infliximab and methotrexate in the treatment of
rheumatoid arthritis. Anti-Tumor Necrosis Factor Trial in Rheumatoid Arthritis with Concomitant Therapy Study
Group. N Engl J Med 2000; 343 (22): 1594-602.
15) National Institute for Health and Clinical Excellence. Guidance on the use of etanercept and infliximab for the
treatment of rheumatoid arthritis. Technology Appraisal No. 36. March 2002. http://www.nice.org.uk/pdf/RA-
16) National Rheumatoid Arthritis Society (NRAS). The National Rheumatoid Arthritis Society Survey, April 2003
17) Norfolk and Norwich University Hospital. Vial Optimisation: Maximising the use of scarce resources
(Unpublished, personal communication. Submitted in confidence).
18) Olszynski WP; Davison KS. Treatment compliance of rheumatoid arthritis patients receiving infliximab and
etanercept: the Saskatchewan experience. Ann. Congress European League Against Rheumatism (EULAR),
Berlin, Germany, June 9-12, 2004, Abstract No. FRI0141.
19) Schering-Plough Ltd. Infliximab data from the British Society for Rheumatology Biologics Register.
Supplement to NICE submission: Remicade® in the treatment of rheumatoid arthritis in England and Wales,
20) Wolfe F. The prognosis of rheumatoid arthritis: assessment of disease activity and disease severity in the
clinic. Am J Med 1997;103 (6A): 12S-18S.
Reference: Schering-Plough Page 12 of 12