APPENDIX 2
Memorandum by Sheila M Bird (NC 11)
Individual submission from first statistician-member
of NICE's Appraisal Committee. For potential conflicts of interest,
please see publicly available Declarations of Interest re NICE.
I preface my remarks by paying tribute to the
tremendous enthusiasm, dedicated hard work and intellectual openness
of the professional and administrative staff at NICE; and to Professor
Barnett's splendid, indefatigable chairmanship of the Appraisal
Committee. Members commit time, energy and thought because rigorous
appraisal of the cost-effectiveness of technologies is an emerging,
exciting, applicable science. Methodologies are being researched
and importantly refined, including by the pharmaceutical industry,
in response to the challenges that NICE presents.
Appraising the cost-effectiveness of future
technologies demands that attention is paid now, both by sponsors
and by the National Health Service, to the design and data requirements
of cost-effectiveness information systems [C-EIS]. Typically,
C-EIS goes beyond randomised controlled trials of short-term efficacy:
it shines a new cost-effectiveness spotlight on the statistical
and economic modelling of disease progression. Critically, ethical
consideration needs to be given to patients' informed consent
to data acquisition for C-EIS, which may entail the salient use
of disease registry of NHS costs data or access to anonymized
individual patient data from industry or other pivotal randomised
trials that predate the C-EIS conception. My recommendations are:
Recommendation 1: Invocation of Medicines Act
When research is commissioned by NICE to improve
the cost-effectiveness evidence base on which the Appraisal Committee
will make its decision, NICE should be able to invoke the powers
of the Medicines Act so that the research team has access to anonymized
individual patient data from pivotal randomised controlled trials.
NICE-commissioned research into cost-effectiveness of disease-modifying
treatment of multiple sclerosis was not accorded full access by
all pharmaceutical companies. Consideration should be given to
more general circumstances than NICE-commissioned research when
NICE may invoke disclosure of data under powers similar to the
Medicines Act.
Recommendation 2: NICE forum for sponsors to pre-discuss
Cost-Effectiveness Information System [C-EIS]
Appraisal of cost-effectiveness uses information
sources beyond the pivotal randomised controlled trials on which
licensing was approved on the basis of quality, safety and efficacy
in the short-term. In particular, costs, utilities and longer-term
effectiveness may be assessed outwith the original trials, and
the latter may be sensitive to the chosen statistical model for
disease progression as well as to the choice of data sources by
which to estimate the model's parameters.
Model and data uncertainty, as outlined, mean
that the challenge of appraising cost-effectiveness is much greater
than for efficacy. The methodologies, including Bayesian model
criticism, for doing so are only just being developed for practical
application. Pharmaceutical companies, many of which have responded
impressively to the challenge, should therefore have a forum at
NICE for pre-discussion of the cost-effectiveness model that they
intend to use and of the data sources by which they will derive
parameter estimates for their proposed model: in short, a forum
for pre-discussion of their proposed cost-effectiveness information
system [C-EIS], just as USA's Food and Drugs Administration engages
in pre-discussion of the design of a sponsor's pivotal randomised
controlled trials.
Recommendation 3: Operational criteria for selection
of technologies for appraisal at NICE
The least transparent, and therefore potentially
the least scientifically defendable, aspect of NICE Appraisal
is how the selected technologies are chosen. Scientifically-sound
publicly-disseminated operational criteria should be developed
for the sifting of technologies for NICE Appraisal. Each proposed
topic should be documented [for example on NICE website] as to:
date of proposal, proposer, date of classification, classifier,
actual classification on each of the operational criteria for
sifting, date of selection decision, triage decision [NICE Appraisal
recommended, pended, rejected].
I suggest that, once defined, the new objective
procedure should be applied in retrospect to the last 500 proposed
topics that were filtered informally by health departments and
Ministers. Moreover, post-appraisal classification by NICE on
each of the operational criteria should be compared with the departments'
publicised pre-appraisal classification for technologies which
were subject to NICE Appraisal so that feedback improves the operational
criteria or their implementation, as necessary.
Recommendation 4: Alternative to NICE Appraisal
Since members of Appraisal Committees are unpaid,
their and their employers' generosity of time and effort should
not be squandered on inappropriately selected topics, but used
to the maximum benefit of patients and staff in the National Health
Service.
Alternative to NICE Appraisal should therefore
be available to Ministers when, for example, an appraisal topic
is identified with high NHS budget impact or pre-appraisal guesstimated
high cost but low effectiveness on which there are not even any
properly conducted, relevant randomised trials of efficacy. In
addition, the attention of UK Royal Colleges and of the Medicines
Control Agency, or its European counterpart, should be drawn to
such topics if licensing approval in the last decade has been
given on non-randomised evidence.
Recommendation 5: Enhanced data quality by the
date of NICE re-appraisal
If Appraisal Committee has made recommendations
for an enhanced cost-effectiveness evidence-base, then consideration
needs to be given by Ministers to whether re-appraisal of cost-effectivess
should proceed (or alternative mechanism be invoked) if there
has been no enhancement by the sponsor or others in the evidence-base
by the re-appraisal date.
January 2002
| 
