[Ip-health] For Professor DiMasi, on the size, costs of trials

Jamie Love james.love at keionline.org
Tue Mar 15 06:01:44 PDT 2016

Joe, I wanted to elaborate on why more transparency of the data on the
costs of trials is needed for people to evaluate your study.

1.  The unique project level data in your paper is from clinical trials,
and it drives the entire estimate.

Your mean estimate of Phase 1, 2 and 3 trial costs is $339.3 million (Table
2), which for short I'll call TRIALC

Your other estimates can be described as follows:

* Adjusted for risk of failure trial costs = TRIALC x 2.844 =  $965 million

* Adjusted for risk and capital cost  =  TRIALC x 2.844 x 4.3 = $1,460

* Pre-clinical costs =  TRIALC x 2.844 x .4456 = $430 million

* Pre-clinical costs adjusted for capital costs = TRIALC x 2.844 x .4456 x
2.55 = $1,098 million.

And given the cumulative multipliers, this can be expressed as:

** Total cost = TRIALC x 7.627 = $2.588


Feel free to fine tune this, but as far as I can tell, everything flows
from the estimate of the trial costs.

2.    The $339.3 figure for trials, before adjustments for risk or capital
costs, seems high, and the ratio of the mean to median seems low.  But
knowing some details about the numbers would be helpful.  That is why we
wanted to see the number of patients and the cost per patient numbers,
since those can be compared to drugs where data is widely available.

For example, for Xtandi, the number of patients in the trials cited in the
FDA review for the approval on the lead indication were fairly small, and
the overall cost was probably much lower than your estimate.  The press
release on the drug's approval mentioned trials involving 1,119 patients,
and the FDA medical review referred to 1,426 patients.   PhRMA recently
estimated the average cost of oncology trials at $59,500 per patient.
1426 x 59500 =  $88.847 million.  That would be just 26 percent of the
average cited in your study.   Moreover, the Xtandi approval involved more
patients that did the average cancer drug over the five year period
beginning in 2010  -- most of the cancer drugs had fewer patients, and
probably lower costs.  So, does your study offer useful estimates for
cancer drugs, a question that UACT posed earlier.

PhRMA estimates the average cost of trials for infectious diseases to be
$16,500 per patients.  The FDA medical review for the HIV drug rilpivirine
referred to trials with 1747 patients.   1746 x 16500 = $28.8 million,
which is just 8.5 percent of your average.

I could go on and provide all of the trials from 2010 to make my point, but
overall, both the average and the median products from all approvals would
appear to be far lower in cost, for the approval of the lead indication on
the drug, and the figures that drive your study, and considerably more
skewed in terms of the distribution, which raises questions about your

Of course, I believe you have done something other than estimate the costs
of R&D for the lead indication, but given the lack of transparency on the
data that is driving your estimate, it is very challenging to understand
what you claim to estimate, and how your sample compares to the average
drug approved by the FDA in a given year.



James Love.  Knowledge Ecology International
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