[Ip-health] The variables in Dr. Peter Bach's Drug Abacus

Jamie Love james.love at keionline.org
Fri Jun 19 16:46:52 PDT 2015


On Fri, Jun 19, 2015 at 4:29 PM, Joel Lexchin <jlexchin at yorku.ca> wrote:

> Jamie:
>
> I have to agree with Ruth. Novelty is only valuable if it leads to better
> therapeutic outcomes. The glitazones - rosiglitazone (Avandia) and
> pioglitazone (Actos) work through a novel mechanism and are public health
> disasters. The drugs that you cite worked through novel mechanisms but had
> public health benefits. Usually by the end of a phase III trial you know
> how efficacious the drug will be although safety is often only really
> determined after the drug is on the market and large numbers of people have
> used it. Larger number of people in clinical trials may be a proxy for R&D
> costs but it’s also a proxy for drugs having less additional therapeutic
> benefit over what’s already on the market. One of the reasons for large
> clinical trials is the need to achieve statistical significance - the
> larger the trial the easier it is to get a p value less than 0.05. If a
> drug is really valuable then you don’t need large number of patients to
> show it. (This argument doesn’t apply to trials on drugs for rare diseases
> because in these cases small trials are the norm.)
>


​Dear Joel,

If you are trying to determine if a drug has a therapeutic benefit, then
only the variables dealing with its therapeutic benefits are relevant.
And, in some pricing models, this may be all you need, for example, if you
are basically buying therapeutic benefits, and that's all that counts.   I
understand this line of reasoning.

If the pricing/reward model is, on the other hand, expected to be a more
optimal incentive mechanism, for innovation, only looking at the variables
which measure if the drug had therapeutic benefits are not sufficient, in
my opinion.

There are cases where one wants to consider R&D costs, and I think, as
modeling of incentives advances, this will become more important, then one
has to begin somewhere to estimate R&D costs, and identify proxies for R&D
Costs.   Before getting into the best proxies discussion (you can have few
or many),  I should say why I think R&D costs are going to be relevant in
pricing models.   Drugs for diseases with small populations often have very
high prices, and these are only tolerated because implicitly, people think
you have to spread a high fixed cost over fewer patients.    Many of the
complaints over sofosbuvir pricing are really about the opposite situation,
the fixed R&D costs are being spread over a larger population, but the
prices are still high, because the "value" of the medicine, for the sickest
patients, is high, by traditional measures.  (not true for all infected
persons, but true for the small percentage that is very sick).

There are also IPR incentives to induce testing drugs for pediatric uses,
or to test drugs on new indications, or to have additional safety and
efficacy data.  In these cases, there is the policy issue of how rich does
the incentive to have be to induce someone to made an investment in the
desired tests?

Our own research does indeed show what you suggest above, that drugs that
are medically important generally have smaller trials than drugs that have
fewer medical benefits (over existing drugs), probably for the reasons you
suggest.   One thing we have little data on, but we assume, is that drugs
that use a new mechanism involve higher risks, at least in pre-clinical
research (an area with the least data available).   I think it is
understandable that researchers at Sloan Kettering see benefits in
establishing new mechanisms to treat diseases, and I assume this is because
the knowledge about the new mechanisms will deepen our scientific knowledge
about the disease and the possible ways to treat the disease, and create
the possibility of further types of follow on drugs using the new
mechanism, including drugs that are better than the early ones.   That
would justify, from an R&D incentive perspective, a higher value for the
incentive, even if the drug itself is not better, or more to the point, all
other things being equal.

People will came at the pricing models from different perspectives, for
very legitimate reasons.  An entity with limited resources that is seeking
to get as much medical benefit as possible for procurement dollars, given
what is on the market, could easily justify looking at the therapeutic
benefits, and calling it a day.

Some actors, and perhaps society as a whole, might also want to at least
model how one could restructure incentives to stimulate the most optimal
investments in R&D, for a given amount of money devoted to financing
incentives.

Our current research interests including looking at this last question, and
here, we think having data and a theory of how to use data on R&D costs
will be important, not to the exclusion of data on health benefits, which
of course are a first order concern, but as a complement.   I'm not saying
you don't also think so, but I'm just trying explain our focus on this
right now.

As an aside, as regards R&D data, the legislation in California and other
states will be much more useful if the data is as dis-aggregated as
possible, with audited costs for each individual trial, and data on the
timing of the trials.  The data available on the economics of drug
development, and also the revenues from drug sales, should be a good as
possible, and as detailed as possible, for good policy analysis, just as
the medical data from trials should be as transparent and detailed as
possible, while protecting patient privacy (and only patient privacy).

Jamie





http://www.keionline.org/donate.html
KEI DC tel: +1.202.332.2670, US Mobile: +1.202.361.3040, Geneva Mobile:
+41.76.413.6584, twitter.com/jamie_love



More information about the Ip-health mailing list