How to choose a real winner in the Covid-19 vaccine race


This article is part of a series in which key commentators explore innovative scientific, medical and epidemiological approaches to the Covid-19 pandemic.

The writer directs JW Scannell Analytics

Suppose a bank decides to choose the winners for its loan portfolio by lending only to companies with visionary CEOs and disruptive strategies. And then it turned out that the only visionary companies that were disruptive enough to borrow money were the ice cream makers. The bank might realize that it had made an involuntary bet on the weather. It should cover the risk with subprime umbrella bonds.

The world is now choosing the Covid-19 vaccines, which we hope will all be winners. Vaccine development normally takes a decade. But in a matter of months, what was unprecedented has become commonplace in terms of speed, regulation, spending and philanthropy. Most optimistic experts now believe there may be a deployable coronavirus vaccine this year (others are more pessimistic). There are at least 90 projects, with around 10 leaders. These 10 were either part of the first human trials, or have the support of one of the major pharmaceutical companies.

Agencies have worked hard to bring order to creative chaos. These include the Coalition for Epidemic Preparedness Innovations, the Wellcome Trust, the Gates Foundation, the World Health Organization, several national governments and their agencies, and various associations in the pharmaceutical industry. But in this race, as in the ice cream example, inadvertent bets can be made. Portfolio risk analysis and hedging should therefore start early.

A central idea is the diversification or the guarantee that the vaccine candidates do not all share the same risks. Suppose we have a bunch of mediocre vaccines, each of which is barely 10% likely to work, but they were likely to fail for entirely different reasons. The chance of not working is 0.9 brought to the number of candidates. If we put 20 to the test, there is almost a 90% chance of at least one success. Compare that to a set of great vaccine candidates, each 40% likely to work, but all will likely fail for the same reason. Here we would be stuck with a 60% chance of failure no matter how many we tried.

The important, but counter-intuitive message is that when you have a large number of vaccine candidates for the same disease – as is the case today – it is more important to estimate and manage correlations in the portfolio than adding another “winner”.

If we look at the precursors of Covid-19, individual rational choices may have concentrated the risk in terms of vaccine platforms used and specific parts of targeted viral machines. Guidelines to help select candidate vaccines for clinical development might end up prioritizing the same thing. An open movement towards diversity is necessary, even if individual projects seem less attractive.

Another major concern is the public’s fear of vaccines. If the first Covid-19 vaccine were to emerge with a safety concern, it could hinder the adoption of follow-up vaccines. At the very least, this would complicate subsequent testing and delay launches. We may insist that we be even more confident than usual about the safety of the first vaccine before mass deployment.

Diversification is also undermined by therapeutic nationalism, whether in a weak form (such as the well-intentioned support of national winners) or a strong form (a bias to supply the internal market). Governments must increase, diversify and cover national efforts.

It’s cruel to add to the to-do list. But here are three propositions. First, vaccine sponsors should adopt a formal portfolio analysis. Second, governments should consider the global vaccine portfolio when deciding which national projects to defend. Third, Cepi or WHO should become risk managers of last resort. They should invest in umbrellas in case it rains on what others are doing.

Vladimir Shnaydman from ORBee consulting also contributed to this piece


Please enter your comment!
Please enter your name here