NICE and value based pricing: what do we know?

Part 1

In the debate on value-based pricing, one issue that the authors
always address either explicitly or implicitly is the comparison between
value-based pricing versus "traditional methods" of price determination
(Collier 2010; Coombes 2010; Delamothe 2010). These latter methods share a
common characteristic in that the value-based approach is not employed for
price determination; if one moves from a negative definition of the term
to a more positive definition, the term "traditional methods" implies that
an empiric approach for price determination is used in which the main
determinant is the need to keep an international homogeneity in prices and
to consequently discourage the parallel import between different
countries.

One feature of the Italian pharmaceutical market is that the
"universalistic" coverage provided by the National Health System has led
to the reimbursement of all innovative agents approved by EMA. Up to 2009,
value-based methods of price determination have not been used in Italy,
and so the "real" prices of innovative drugs approved in this country are
an excellent and systematic reference to document the results that
"traditional methods" can produce in terms of drug prices. In addition,
since one experience has been carried out in Italy in which the value-
based prices of new agents were systematically determined (Santarlasci et
al. 2008), the information available in Italy offers a unique chance to
compare real (or "traditional") prices and value-based prices with one
another.

The experience by Santarlasci et al. (2008) has led to the comparison
between real vs value-based prices for 122 innovative interventions (data
from the website www.osservatorioinnovazione.org accessed on 30 November
2010). After excluding the subgroup of drugs or devices for which
superiority was not demonstrated and/or price was still unknown and/or the
analysis could not be made, the remaining innovative treatments were 77,
of which 60 belonged to the oncology area while 8 were non-oncologic
drugs, 8 were medical devices and 1 was a decontamination procedure.

The median ratio of real price vs value-based price was 1.15 in the
overall group of 60 oncologic agents. In the other subgroups, that were
less informative due to their small number, this median was 0.37 for the 8
non-oncological drugs and 1.39 for the 8 devices.

Among the 60 oncological drugs, 37 met the definition of
biotechnology drugs or "small molecules" designed as target therapies. The
analysis of these 37 cases showed a median ratio between real price and
value-based price of 2.15 (see Table 1 for details). This latter result is
of particular interest because of the large number of drugs included in
the analysis and also because it refers to the therapeutic area in which
most pharmacological innovations are being developed.

References

- Collier J. Drug pricing, NICE, and the PPRS: government gets it
right. BMJ 2010; 341:c6378

-Coombes R. Briefing: NICE and value based pricing: what do we know?
BMJ 341:doi:10.1136/bmj.c6390

-Delamothe T. A future for NICE, and the world's poorest children.
BMJ 2010; 341:c6422

-Santarlasci B, Burchini G, Simbula S, Trippoli S, Messori A.
Progetto 'Osservatorio sull'innovazione': schede web sui prodotti
innovativi comprensive di analisi economica semplificata. Giornale
italiano di Farmacia clinica 2008;22:24-30.

Table 1. This table can be seen either at the address
http://www.monitoraggiterapeutici.org/supplements/fadda-rr-dec2010.htm or
in the second part of this Rapid Response.

Dear Dr Prieto

You have offered a method which you call the "Complex Number" method
and claim that it is superior to the standard way of calculating QALYs as
utility multiplied by time. As your paper (1) makes clear, all that is
really involved in your method is a non-linear transformation of the
utility scale before multiplying by time.

It needs to be remembered that utilities are not intrinsic properties
of health states, but are obtained from a valuation exercise in which some
compromise is made between the required properties of the resulting
values. Methods of eliciting utilities attempt to ensure, for example,
that the difference between utilities of 0.1 and 0.3 is the same as the
difference between utilities of 0.6 and 0.8, and would give a 0.2 QALY
gain for a year in the better state in each case.

Applying your non-linear transformation would give a QALY gain of
0.0276 for a year's improvement from a utility of 0.1 to 0.3, but a QALY
gain of 0.0809 for a year's improvement from 0.6 to 0.8. This is a serious
distortion of the elicitation process. Your transformation also gives a
maximum gain of less than 0.3 QALY for a year's improvement between any
two states better than dead. As you state explicitly in your paper, your
method gives a minimum gain of over 0.7 QALY for a year's extended life in
any health state better than dead, however poor the state: I do not see
this as a desirable property of a method.

You claim without any supporting evidence that the zero on the
utility scale is arbitrary and that this is not debatable. In my previous
response, I gave a reason in favour of the view that the zero is entirely
natural. Indeed, it seems to me that the alternative contention that the
zero point is arbitrary is untenable. In my previous response, I leant
over backwards to try to show that this did not matter. Strictly, my
argument only applies if there is no mortality difference between the two
options being compared, and I now acknowledge that limitation.

You have raised the issue of states worse than death. There is
certainly no perfect method of handling those. From my point of view, if
some of the people surveyed rate a state as better than death and others
rate it as worse than death then the only question to ask is how to
average their views.

Your paper (1) does not make clear how you would calculate QALYs from
negative utilities. Your formula would give the same results for a utility
of minus 0.1 as for a utility of plus 0.1 and more QALYs for a year at a
utility of minus 0.2 than for a utility of minus 0.1. I am sure this
cannot be your true intention.

1. Prieto L, Sacristan JA. Problems and solutions in calculating
quality-adjusted life years (QALYs). Health Qual Life Outcomes 2003;1:80.

It can be argued that the zero point for utilities is not arbitrary,
but a natural zero, as no further QALYs can be accumulated after death.

However, even if the utility scale is regarded as only having
interval properties, it is perfectly permissible to multiply values on the
utility scale by values on a ratio scale (time) to produce answers (QALYs)
which would then be regarded as being on an interval scale.

It is then also perfectly acceptable to calculate the difference
between expected QALYs for a patient group under two different management
strategies (one of which could be "no treatment"). Differences between
values on an interval scale themselves lie on a ratio scale and can be
used for decision making.

This distinction between QALYs achieved under one treatment strategy
(which may only be on an interval scale) and QALYs gained by using one
treatment strategy rather than another (which are certainly on a ratio
scale) is fundamental to cost-effectiveness analysis.

It is no help to decision makers to say that a particular measure
should not be used. If QALYs are not to be used, then an alternative
method of decision making must be proposed, and it must be shown that the
balance of advantages and disadvantages favours that alternative method.