APPENDIX 9
Memorandum by the King's Fund (NC 46)
The King's Fund supports the use of explicit
priority setting within the NHS and is broadly supportive of the
objectives and processes of NICE. Rationing within the NHS—even
at vastly higher levels of expenditure—is unavoidable, and
explicit approaches have the potential to facilitate consistent,
transparent and accountable decisions about resource allocation.
In the following submission we outline six aspects[14]
of NICE's processes which we consider could be strengthened or
improved to better meet its objectives. In each case, we briefly
outline the relevant issues and offer a specific recommendation
for change. Analysts from the King's Fund are willing to provide
oral evidence if required.
SUMMARY OF
RECOMMENDATIONS:
1. That NICE retain its NHS cost perspective,
but accompany this with an impact report of the resource implications
for public and private groups and individuals.
2. That NICE explicitly justify its decision
threshold cost per QALY gained, and state what tradeoffs between
QALY maximisation and other objectives (such as minimisation of
unequal health status) it is prepared to accept.
3. That guidelines for evidence on the collation
and reporting of patient experience be accompanied with clear
procedures for how this evidence will be used to make comparisons
between patient groups.
4. NICE should develop systematic ways to
choose which services and treatments are to be appraised rather
than responding mainly to new technologies.
5. That NICE work with manufacturers of
pharmaceutical products to ensure that appropriate measures of
quality of life are routinely included in clinical trials.
6. That local variations in marginal cost
and benefit, and cost effectiveness in implementation, be explored
as to their implications for technical appraisal and national
guidance.
1. Costs
NICE appraisals are undertaken from the perspective
of the NHS. This means that resource use associated with conditions
and their treatment is restricted to that of the NHS. This raises
a fundamental question: whose budget should we be concerned with?
The NHS? The Public Sector? Patients? Or the whole economy?
There is no technically correct view about this.
Focussing solely on publicly-funded healthcare costs and ignoring
costs falling on patients or the wider public, can mean that appraisals
favour interventions which shift costs away from the NHS onto
private individuals or the rest of the public sector. The desirability
of an all-encompassing (social) perspective is emphasised by many
prominent writers in economic evaluation.
But a wider perspective on costs would require
NICE to extend its remit to optimising resource use in the public
sector generally, not just health care, and indeed in the economy
as a whole. Decisions made from this broader perspective might
maximise society's welfare, but would not maximise health gain
from the NHS budget. Including non-NHS costs would create still
more uncertainty over what is an appropriate cost-effectiveness
threshold (see 2 below), as resource use outside the healthcare
budget is more difficult to identify, measure and value. Moreover,
it would require a higher threshold; some conditions imposing
low non-NHS costs might no longer be regarded as cost-effective.
Recommendation 1: That NICE retain its
NHS cost perspective, but accompany this with an impact report
of the resource implications for public and private groups and
individuals.
2. The cost-effectiveness threshold
Cost effectiveness evidence has no "absolute"
interpretation—whether or not a treatment with a given cost
per QALY gained is considered to give good value for money or
not relies on comparison of this evidence with that for other
services that are already funded (but this assumes currently funded
services to have been rationally chosen and to be good value for
money), or against a benchmark known as a "threshold".
Guidelines issued by NICE to date imply a threshold
in the range of £20,000 to £30,000. Statements made
at the AGM of NICE suggest a de facto threshold of £30,000.
However, cost-effectiveness is not the only factor considered
in issuing guidance. For example, NICE decided in favour of Riluzole
for motor neurone disease, despite a cost per QALY gained of between
£34,000 and £43,500, stating that the committee took
account of "the severity and relatively short life span"
of patients and "the value patients attached" to the
extension in survival.
The threshold is a way of representing the opportunity
cost of treatments "at the margin"—it represents
society's willingness to pay for an extra QALY. But because we
lack information on the cost per QALY gained for all possible
health care interventions, the threshold represents an estimate
of society's willingness to pay.
NICE has a responsibility explicitly to justify
the threshold that appears to have emerged as its decision rule.
For the sake of consistency and transparency, NICE should also
explicitly state which factors other than value for money—for
example, reductions in health inequality—are considered
important and how they are incorporated into decisions. The choice
and use of the threshold are critical in decision making, and
making these explicit would enable more transparency and better
public debate on these matters.
Recommendation 2: That NICE explicitly
justify its decision threshold cost per QALY gained, and state
what tradeoffs between QALY maximisation and other objectives
(such as minimisation of unequal health status) it is prepared
to accept.
3. Patient experience
Both the DoH and NICE have declared their intention
to seek patient (or public) input to their decisions. It will
be important to state clearly the objective of doing so and to
establish guidelines for collating and presenting such evidence
so that it facilitates NICE's overall aims.
There are risks inherent in relying on individual
patient or lobby group representations, including the selective
nature of such evidence and the possibility of bias against health
services where individual patients cannot be identified (eg, public
health strategies) or where lobby groups have not formed. Systematic
recording of patient experiences, using qualitative research methods,
offers an alternative. However, the key point here is that explicit
priority setting requires evidence to be weighed up across patient
groups and treatments which represent potential claims on resources.
A decision to recommend one sort of treatment, which meets one
set of patient needs, has the opportunity cost of denying other
treatments, which meet other patients' needs. Evidence of any
kind—whether cost per QALY gained, or patient experience—is
useful only if it facilitates such comparisons; this is the essential
task faced by NICE. Unfortunately, qualitative evidence is hard
to compare in this manner. What is done with evidence on patient
experience is, therefore, as important as the method used to generate
it.
A central concern for many patient groups seems
to be the inadequacy of quality of life measurement and valuation
techniques used in technical appraisals. There may also be other
concerns—for example, that the way in which health states
have been modelled, such as their duration, transition probabilities
and side-effects from treatment, fails to capture the reality
of patients' experiences. Evidence about patient experience might
provide a basis for judging the reliability and reasonableness
of technical appraisal results. This will require an extra step
in the process: rather than just providing a patient impact assessment
report, it needs to feed into and inform the appraisals. If this
is not done, even the most carefully analysed qualitative evidence
is likely to become sidelined in the decision making process,
since it will not in itself facilitate between-group comparisons
of need or ability to benefit.
Recommendation 3: That guidelines for
evidence on the collation and reporting of patient experience
be accompanied with clear procedures for how this evidence will
be used to make comparisons between patient groups.
4. The appraisal agenda
Evidence-based priority setting faces the challenge
of how to deal with missing evidence—this affects not only
how but also which things are appraised. For example, bias in
the selection of services and products towards those that have
evidence from randomised control trials can distort decision-making.
It makes it more likely that priority setting focuses on the ranking
of relatively new treatments, while existing services, which are
less likely to have been subject to trials, remain outside the
scope of assessments, while continuing to make a claim on resources.
Under these circumstances, new services might fail to clear the
efficiency threshold and be denied funding, while "old"
services remain embedded which may be less cost-effective. Appraisal
processes need to be accompanied by clear methods, based on economic
evaluation principles, to identify which services get appraised.
A variant of Programme Budgeting Marginal Analysis could provide
a way forward.
Recommendation 4: NICE should develop
systematic ways to choose which services and treatments are to
be appraised rather than responding mainly to new technologies.
5. Quality of life measurement
The lack of evidence on the impact of treatments
on quality of life can act to limit the scope and the veracity
of analyses of cost per QALY gained. In order for a pharmaceutical
manufacturer to obtain a licence for a product, it must meet certain
regulatory criteria, which involves their providing evidence on
safety, efficacy, and production standards for the product from
their trials. There is no requirement to collect quality of life
data, and no guidelines are provided for the type of data (eg,
the instruments used to measure quality of life) that should be
collected. Incorporating quality of life measurements into clinical
trials imposes significant additional development costs on manufacturers—in
order to produce this evidence they need to face incentives to
do so.
There are two options. One is to mandate the
inclusion of QoL measures into clinical trials—a move likely
be unpopular with the industry (and potentially inefficient since
many drugs do not proceed past early phases anyway). A second
option is for manufacturers to be encouraged to include QoL measures
routinely into trials—and for this to be accompanied by
clear guidelines as to the type of measures that are expected
to facilitate NICE appraisals. We would suggest that these should
include a patient-experienced, condition specific measure, as
well as generic measures for which health state valuations (of
the type used in estimation of QALYs) are available.
Recommendation 5: that NICE work with
manufacturers of pharmaceutical products to ensure that appropriate
measures of quality of life are routinely included in clinical
trials.
6. Localism versus centralism and allocative
efficiency
One of the purposes of NICE is to provide clear,
nationally consistent guidance on resource use. However, another
key goal is to provide guidance that is locally owned and acted
on in the right way. These two objectives may not be easily reconciled.
In practise, NICE technical appraisals are based upon national,
average data. Yet the basis for economic evaluation is in incremental
analysis: allocative efficiency relates to marginal costs and
marginal QALYs gained. There are a number of reasons why there
may be differences between national and local costs and benefits.
First, where local marginal costs are different
from national average costs—due to different levels of throughput,
spare capacity, and economies of scale and scope in provision—efficient
service levels may also differ.
Second, whether a treatment demonstrated to
be cost effective at a national level is so locally depends in
practice upon the way in which NICE's Guidance is implemented
locally. For example, Guidance issued in favour of Riluzole may
be interpreted in one area as an additional service for patients
with MND (and resources taken from other treatment areas); in
others, Riluzole may be funded but support services for MND patients
withdrawn. Thus the budget constraint, net budgetary impact, opportunity
cost and incremental benefit from that treatment are potentially
different in each setting than estimated nationally.
While inter-regional variations in service availability
probably do reflect non-optimal (and inequitable) patterns of
provision, neither is it guaranteed that identical services are
optimal. Thus there needs to be provision for local consideration
of costs and benefits (especially given the newly mandatory status
of NICE's guidance) and the extent to which national average data
is a good reflection of the conditions prevailing locally.
Recommendation 6: That local variations
in marginal costs and benefit, and cost effectiveness in local
implementation, be explored as to their implications for technical
appraisal and national Guidance.
CONCLUDING STATEMENT:
The processes employed by NICE to assess the
effectiveness and cost-effectiveness of spending options facing
the NHS are based on sound principles. We have outlined a number
of means by which those processes could be strengthened and made
more explicit in order to better meet the stated objectives of
NICE.
Many of the concerns voiced publicly by stakeholder
groups regarding NICE appear to focus on the use of economic evaluation
and QALYs in particular. Cost-effectiveness analysis of the kind
used by NICE is not without its problems, and provides just one
kind of information (ie, efficiency, defined in terms of maximising
QALYs), which must somehow be balanced against other, possibly
conflicting, health service objectives (such as the minimisation
of health inequalities) and principles (such as the rejection
of "ageism" in health services). Nevertheless, the heart
of the rationing problem is that devoting resources to one treatment
can only be achieved by denying the use of those resources to
fund other treatments. Evaluation of cost per QALY gained represents
by far the best means of assessing value for money across different
conditions and treatments.
December 2001
14 Part of this Evidence draws from "Patients'
views of explicit priority setting: what are the implications
for NICE?", Devlin N, Appleby A, Parkin D (2001) King's Fund. Back
| 
