Transcript Valuing Trial Designs from a Pharmaceutical Perspective using

Valuing Trial Designs from a Pharmaceutical
Perspective using Value Based Pricing (VBP)
Penny Watson1, Alan Brennan1
1Health
Economics and Decision Science, ScHARR, University of Sheffield, UK.,
Introduction
Conclusion
Expected Net Benefit of Sample information (ENBS) can be useful in deciding which
data collection strategy is optimal 1. The traditional ENBS is not compatible with drug
development in the pharmaceutical industry because ,
o
Traditional ENBS value trials according to the expected benefits to society,
o
Traditional ENBS assumes the price of the intervention is fixed.
We aimed to evaluate trial designs for Systemic Lupus Erythematosus (SLE) drugs.
We have illustrated how ENBS can be adapted to value clinical trials in the
pharmaceutical industry using expected VBP to integrate price uncertainty into the
decision criteria. Our analyses indicated that larger sample sizes are more efficient
than longer trials in SLE.
This very simple example took 5 days to generate 10,000 sets of trial results. The
analyses can be very time-consuming to run for complex economic models.
Methods
Results
We developed a simple CE model for SLE in which costs
and QALYs were estimated analytically conditional on
average lifetime disease activity, average lifetime organ
damage and mortality.
Profit forecast (PF)was estimated from,
PFxn d  VBP ( | X nd )tkhs
 =CE model parameters, X=data, t=drug maintenance,
500
Increasing sample size and duration of follow-up
increases the expected VBP. However, increasing sample
size with duration of follow-up of 3 years has no impact on
the Expected Value Based Price (Table i).
400
Table i: Expected VBP, Profit Forecast, and Trial Costs
Follow-up
300
1 year
2 years
3 years
VBP Profit Cost VBP Profit Cost VBP Profit Cost
200
n
100
VBP described the maximum price given the willingness to
pay threshold of the reimbursement authority (£30,000).
VBP was zero if the trial endpoint was not statistically
significant, or if the VBP was less than the minimum
acceptable price of £1000 per year.
d=1
d=2
d=3
(£)
(million £)
(£)
(million (£)
(£)
(million £)
100
£786 £461 £1.1 £865 £458 £1.15 £892 £420 £1.2
500
£902 £530 £1.5 £905 £478 £1.75 £905 £425 £2.0
1500 £911 £535 £2.5 £911 £481 £3.25 £903 £427 £4.0
0
We updated the CE model with trial data using the Brennan
and Karroubi Bayesian Approximation method 2-3.
Net Benefit millions £
We sampled 10,000 trial datasets (X) for nine trial designs
(n=100, n=500, n=1500, d=1, d=2, d=3) in which disease
activity, organ damage and treatment withdrawal were
collected.
Figure i:Net Benefit of Trial Design
n=100
n=500
n=1500
k=incidence, h=drug life horizon, s=market share
Contact
Contact: Penny Watson
Postal address: ScHARR, Regents Court, 30 Regent Street, Sheffield S1 4DA, United Kingdom.
Email: [email protected] Website: www.shef.ac.uk/heds
n=100
n=500
n=1500
Trial design
n=100
n=500
n=1500
The trial design with n = 1500 and d=1 years has the
highest Expected Net Benefit (Figure i). Trials with larger
sample size have greater Net Benefit. The duration of
follow-up is negatively associated with Net Benefit due to
the increased costs of the trials, and a shorter life-horizon
of the drug.
References
1. Claxton K. The irrelevance of inference: a decision-making approach to the stochastic evaluation of health care technologies. Journal of Health
Economics 1999; 18(3):341-364.
2. Brennan A, Kharroubi SA. Efficient computation of partial expected value of sample information using Bayesian approximation. Journal of
Health Economics 2007; 26(1):122-14
3. Brennan A, Kharroubi SA. Expected value of sample information for Weibull survival data. Health Economics 2007; 16(11):1205-1225.