Randomized clinical trials and RA cost-effectiveness analysis: is there tarnish on the gold standard?

[Guest guest]

Rheumawire

Jan 26, 2004

Randomized clinical trials and RA cost-effectiveness analysis: is there

tarnish on the gold standard?

Wichita, KS and Berkshire, UK - Cost-effectiveness analyses arguing that

expensive new drugs for rheumatoid arthritis (RA) are worth their high

price tag have become commonplace in the marketing of these new

products. Most such analyses rely heavily on data from randomized

clinical trials (RCTs), which are then extrapolated to clinical

practice. But these RCT data are so far removed from the world of

ordinary practice that they cannot be used to produce reliable estimates

of the cost-effectiveness of new drugs, according to Dr Fred Wolfe and

colleagues, in an editorial in the January 2004 issue of Rheumatology

[1].

Wolfe heads the observational US National Data Bank for Rheumatic

Diseases (NDB) (Wichita, KS). He argues that the observational data

collected by this bank more accurately reflect what happens in clinical

practice than data accumulated in RCTs, and in the editorial he outlines

a comparison between these 2 types of data sets, which showed up a

9-fold difference in the results.

The inspiration for Wolfe's editorial is an article [2] in the same

issue by Dr Alan Brennan and colleagues at the University of Sheffield,

UK, and at Wyeth Laboratories, Berkshire UK. Those investigators used

RCT data and a cost-effectiveness model to argue that etanercept is

cost-effective when compared with nonbiologic agents for treatment of

adults whose RA is inadequately controlled by at least 2 prior

disease-modifying antirheumatic drugs (DMARDs.)

Wolfe describes the Brennan analysis as " the state of the art of

rheumatology [cost-effectiveness] analyses. " But he argues that even

studies as " transparent, careful, thoughtful, and insightful " as this

are flawed because the patients participating in clinical trials are not

representative of those seen in clinical practice, due to differences in

disease severity.

Wolfe also maintains that differences in the Health Assessment

Questionnaire (HAQ) and ACR20 response data from RCTs compared with

studies based on observational data are so great that they call into

question cost-effectiveness conclusions based primarily on RCTs.

Brennan reported that in the RCTs, patients treated with etanercept who

were classed as ACR responders showed an HAQ improvement of 0.84, and

even those classed as ACR nonresponders showed an improvement of 0.37 on

the HAQ. For comparison, Wolfe and colleagues analyzed data for 785 RA

patients from the NDB who were not receiving etanercept when the HAQ was

first assessed but were receiving it 6 months later and had taken

etanercept for at least 3 months. The main HAQ score in these patients

improved by only 0.08 units. Wolfe et al point out that " the difference

between the RCT and the observational study results (which may be

thought to represent practice results) was in excess of 9-fold. "

" There are several possible explanations for this difference between RCT

results and clinical practice, " Wolfe writes. " The primary explanation

is that RCT data and the NDB data use 2 different timings to examine the

effect of etanercept on HAQ. In RCTs, the test is between a deliberately

tested 'worst' state that was identified by inclusion criteria and some

future time (usually 6, 12, or 24 months). In the NDB analysis shown,

the test is between the patient's usual clinical state 1 to 3 months

before starting etanercept and 3 to 6 months after starting etanercept. "

" If the magnitude of benefits seen in RCTs is not of the magnitude seen

in actual practice, then the cost-effectiveness conclusions based on RCT

data are not valid. This is the main objection we have to this study of

Brennan at all, " Wolfe said.

Wolfe also points out that in clinical practice a patient who would have

been considered a responder according to ACR20 or disease activity score

(DAS) 28 criteria in the clinical-trial setting could be judged a

failure due to unsatisfactory functional and pain status or to an

unacceptable rate of disease progression. " Treatment failure and success

in the RCT model is different from failure and success in actuality, "

Wolfe said.

Perhaps unsurprisingly, Wolfe and colleagues suggest that observational

databases may be more useful for cost-effectiveness analyses in RA than

RCT data by themselves. The conundrum this presents is that

observational data are usually not available until after drug approval.

But Wolfe proposes to solve this problem by having clinicians collect

" simple self-report data at all times " so that a context will be

available should the patient later enter a randomized clinical trial.

Cost-effectiveness of etanercept: data used by UK NICE

Brennan's modeling study [2] compared the

cost-effectiveness of a DMARD sequence with etanercept against the same

sequence without etanercept in patients whose RA was inadequately

controlled on trials of at least 2 prior DMARDs. This was examined using

the 6-month trend in HAQ disability scores.

The Brennan model estimated that the cost per

quality-adjusted life year (QALY) gained was £16,330 for the treatment

sequence with etanercept. These conclusions were based on a model that

assumed the population had already failed at least 2 DMARDs, one of

which was methotrexate. Baseline characteristics for the model were

based on a published etanercept monotherapy trial [3]. Mean patient age

was 53 years, 74% of patients were female, and the baseline HAQ was 1.6.

The model was used to simulate the 6-month trend in

HAQ disability scores for 10 000 patients. Drug costs were estimated to

be £30 395 higher in the etanercept sequence but were assumed to be

partly offset by savings of £3506 from reduced levels of disability and

fewer events such as total joint replacement. The etanercept strategy

was estimated to gain an extra 1.66 QALYs. Brennan said that the model

" was an important component " in the decision of the UK National

Institute for Clinical Excellence to accept etanercept as cost-effective

for use in patients who have failed at least 2 DMARDs.

Janis

Sources

1. Wolfe F, Michaud K, Pincus T. Do rheumatology cost-effectiveness

analyses make sense? Rheumatology (Oxford) 2004 Jan; 43(1):4-6.

2. Brennan A, Bansback N, Reynolds A, Conway P. Modelling the

cost-effectiveness of etanercept in adults with rheumatoid arthritis in

the UK. Rheumatology (Oxford) 2004 Jan; 43(1):62-72.

3. Moreland LW, Schiff MH, Baumgartner SW, et al. Etanercept therapy in

rheumatoid arthritis. A randomized, controlled trial. Ann Intern Med

1999; 130:478-486.

I'll tell you where to go!

Mayo Clinic in Rochester

http://www.mayoclinic.org/rochester

s Hopkins Medicine

http://www.hopkinsmedicine.org