Development Efficiency

We can extend the Quick-Kill model to estimate development efficiency - the number of successful compounds in the portfolio for a given development time period, T, and a given amount of development resource.  Development capability can be expressed as the number of new molecular entities (NMEs) that can be processed in parallel - the number of  “pipes” , n, in the development pipeline. 

Once again, let the number of compounds to be screened be N and the proportion of “good” compounds be p.  Let the time taken to develop a “good” compound be t_success and the time taken to kill a “bad” compound be t_kill.  Then the rate at which  “good” compounds are brought to market is given by the number of “good” compounds discovered divided by the time taken to discover them.  Without simplification this rate is given by:

 Np(1-p_fn) / (Np(1-p_fn)t_success + N(1-p)t_kill + Np(p_fn)t_kill )

To calculate the number of good compounds in the portfolio we simply multiply this figure by the number of “pipes”, n, and the period of interest, T.

nTNp(1-p_fn) / (Np(1-p_fn)t_success + N(1-p)t_kill + Np(p_fn)t_kill )

In the Table above Pipeline Performance is expressed as the number of marketable products in the portfolio after 20 years for a Right First Time strategy and a Quick-Kill strategy with various false-negative rates.  In the 20 year period the Right First Time strategy delivers 15.4 molecules.  A fast fail strategy, with an early decision after just one year, delivers more than 21 molecules in the same period.

© Dennis Lendrem, 2011

Quick Win, Fast-Fail

It is not enough to succeed. Others must fail.
Gore Vidal
US author & dramatist (1925 - )

Looking for a needle in a haystack?

Torching the Haystack: Quick Win, Fast Fail


© Dennis Lendrem, 2011

Quick Kill Rides Again

Quick-Kill(TM) Rides Again: More On Quick Win, Fast Fail

© Dennis Lendrem, 2011

Quick Win, Fast-Fail

The overarching benefit of pushing molecules toward early attrition is that it improves the success rate of those that do proceed along the development path.

- Clarke, C. 2010 Quick Win, Fast Fail, International Clinical Trials, Autumn 2010

The Quick-Kill™ Model : Development Savings

In Fail Fast, Fail Cheap, Fail Often we saw that Fast Fail strategies can shorten the expected time to market by clearing failing projects from the development pipeline. Moreover, we saw that this effect is so strong that even if the overall development life cycle for successful products increases dramatically, Fast Fail strategies will outperform strategies based on maximising development speed.

However, it gets even more interesting.

According to the simplest form of the Quick-Kill^TM model, if the probability that a project will be successful is p, the costs of killing a candidate are $C_kill and the development cost for a successful product is $C_success then the expected costs for each successful market launch are

E(Cost) = (¹/_p) C_kill +C_success

Using the Quick-Kill^TM model we can calculate the expected costs per market launch of a development strategy of interest. And if we do this, then one of the findings is that front-loading research and development costs using a Fast Fail strategy produces significant R&D savings.

How does this come about?

For simplicity, let’s assume that as part of a Fast Fail strategy we wish to bring forward a number of tests that would normally be postponed until later in the development life-cycle. In pharmaceuticals, for example, we may wish to introduce an Early-Into-Man strategy with a service formulation knowing full well that if the compound progresses this may entail repeating studies with the final clinical formulation.

Consider the following sets of R&D costs.

In the Right First Time strategy the early development costs to the critical decision point are $20m and the subsequent development costs for successful compounds are just $80m. The overall cost of developing a successful compound is $80m + $20m = $100m.

In the Fast Fail strategy, the key decision is made earlier, after just one year. This is an expensive year, because we have accelerated this phase of development, but the early development costs are still $15m. The initial burn rate for R&D is $15m per annum compared to just $10m per annum for the Right First Time strategy. In addition, the subsequent development costs for successful compounds are $90m - an additional $10m greater than that for the Right First Time strategy. These additional costs are to cover any additional re-work or new studies as a result of the decision to use a service formulation. The overall cost of developing a successful compound is £15m + $90m = $105m.

We can now calculate the expected costs of these two strategies.

Table 1 presents the expected costs of these two strategies – the Fast Fail and the Right First Time strategies – for a range of probabilities that the project is likely to be successful. 

 

For a range of probabilities the Fast Fail strategy outperforms the Right First Time strategy generating signficant R&D savings. Any additional development costs of the Fast Fail strategy are offset by the savings made in killing compounds that would fail later in development.

Copyright Dennis Lendrem, 2011