In the C-Suite, virtually all biopharma companies insist on failing unsuccessful candidates faster and reallocating resources to more promising molecules. This is only rational. But the industry has a dirty little secret. This isn’t the kind of secret that stays within the confines of the C-Suite. It’s the kind that never bubbles up that high.
In practice, a substantial minority of development executives take a different view of early stopping from official doctrine. These executives agree that early stopping is necessary. They disagree about what needs to stop. They want to terminate and often do terminate discussion of including interim reviews and early-stopping rules in a protocol.
This pattern emerges unmistakably if you recommend adaptive techniques to enough companies over time. No sooner has an external design team recommended an interim review with stopping rules than these executives reject the idea without a convincing explanation. Interest in cost savings and the ability to reallocate resources fades away. Vague skepticism and distaste for needless complexity trump the output of thousands of simulations. This pattern is there in big companies and one-product companies alike.
I’ve seen the pattern often enough to realize that the underlying condition needs a name: Early Stopping Blindness (ESB). Many ESB sufferers exhibit emotional commitment to projects that represent years of collective effort. They find failure unthinkable and talk of early termination for futility too unpleasant to contemplate. In more practical and mercenary terms, so long as no one sees data demonstrating futility, the program continues and the project team remains employed. For venture-funded companies, delayed recognition of futility represents a stay of execution.
One intervention shows promise for eradicating ESB: an amnesty program for executives who approve early-stopping rules in the event that an interim review demonstrates futility. Allowing early termination for futility to trigger early termination for executives creates perverse incentives. Rather than striving to learn as much as possible as fast as possible about products in development, incentives drive executives to make decisions that delay the acquisition of knowledge. This contradicts scientific principles and deprives higher management of the ability to make timely, informed decisions. It also maximizes expenditures on futile projects.
Amnesty for early stoppers would eliminate the delay between failing and finding out, enabling rational allocation of R&D budgets. It would also remove an unspoken barrier to wider adoption of adaptive design: fear that earlier acquisition of greater knowledge threatens careers. Adaptive designs have already provided substantial improvements in clinical development, reducing timelines and costs and accelerating knowledge acquisition. The industry will enjoy even greater benefits if we can remove some of the little known human barriers to wider adoption.