Skip to Main Content

We may have a vaccine for Covid-19 by late this fall. By winter, we could have several. And if all goes as planned, each vaccine will be accompanied by data demonstrating that it reduces the risk of developing Covid-19, the clinical syndrome caused by infection with the novel coronavirus.

But we won’t know which of these vaccines is more effective at preventing Covid-19, or have much idea if one or more of them reduce the risk of a person becoming an asymptomatic carrier of the virus. We also won’t know if the duration of immunity is similar between the vaccines, or if the side effects are equally tolerable.

That means someone will need to run a study comparing the vaccines to one another as part of rolling out widespread vaccination. Such a study will ensure that we save the most lives, hurt the fewest, and spend the least.


This clinical trial, which would involve randomly assigning volunteers to different vaccines, would be unprecedented. But so is the situation we find ourselves in. There are at least 40 vaccines against the novel coronavirus being tested in people today. And while each aims to prepare the immune system to stop the coronavirus by simulating a harmless attack from it, the way they achieve this objective is more disparate than we have ever seen.

The range covers vaccines that are simply weakened versions of the coronavirus (such as one from Sinovac) to vaccines that are fragments of the outer shell of the virus (such as one from Novavax). Then there are the Trojan horse vaccines that smuggle scripts of genetic code into the interior of our cells which our bodies unwittingly translate into virus-shaped molecules mimicking the coronavirus (such as ones from Pfizer and BioNTech, Johnson and Johnson, and a collaboration between University of Oxford and AstraZeneca).


It’s hard to predict how well each vaccine will work.  But it’s easy to anticipate that they will not work equally well nor have the same side-effect profiles. We can also anticipate that the various vaccines will differ in their ease of distribution and administration. The vaccines from Pfizer and Moderna need to be distributed frozen, and the companies are testing how much immunity is generated by two injections spaced several weeks apart. Johnson and Johnson’s vaccine only requires refrigeration and is being tested as a single shot.

Here’s where a carefully done comparative study comes in — one I estimate will require around 200,000 volunteers (even more if infection rates decline). That may sound like a big study, but it less than 0.1% of the number of Americans who will ultimately need to be vaccinated.

The study would enroll individuals slated to be vaccinated anyway, and they would be randomly assigned to one of the three to five vaccines that will probably have been approved or authorized for sale by the time the trial begins. No one would receive a placebo, and new vaccines could be included as they become available. This type of study parallels one co-sponsored by the National Institute for Allergy and Infectious Diseases that compared several competing treatments for the Ebola virus a few years ago.

Blinding in this head-to-head trial would not be feasible due to differences in administration schedules. Then again, the ongoing randomized Phase 3 trials are using saline as the placebo shot, which also will probably lead to inadvertent unblinding.

I’m going to assume that one or two vaccines will stand out as the most effective, but it is hard to guess which ones. The study might be able to single out the vaccine that is best for use in the elderly, where minimizing side effects may be more important than provoking the largest immune reaction. It could identify the vaccine that will be best at preventing front-line workers and first responders from becoming unknowing spreaders of the disease. These elements of vaccine performance were flagged as central to any effective vaccine allocation strategy by the National Academy of Medicine.

In an ideal world, this trial would not have been needed. Back in June, the Food and Drug Administration suggested that vaccine developers study similar populations of volunteers and measure similar outcomes so the benefits and risks of the different vaccine products could be compared side by side. But the companies have not followed the guidance that the FDA labeled as “nonbinding.” Now potentially important differences in the respective study populations and outcome definitions may threaten the validity of any post-hoc comparison.

Pharmaceutical companies almost never voluntarily compare their products against their competitors’. The results of such an evaluation could demonstrate, for instance, that Moderna’s vaccine — with an announced price between $32 and $37 — is no better than that from the Serum Institute of India, which the company has pledged to price no higher than $3.

No company wants a comparative study to demonstrate conclusively that its vaccine should be abandoned because it is not as good as available alternatives. Imagine what would happen to the market for Pfizer’s two-dose vaccine if a comparative clinical trial were to show that Johnson and Johnson’s one-dose vaccine is equally effective. Billions of dollars in anticipated revenue could evaporate overnight.

But the traditional preferences of companies should not dictate how we approach vaccine testing in the U.S. Companies usually bear both the cost of their late-stage drug testing and the risk that their products fail to sell. But taxpayers have provided billions in research support to some companies, guaranteed billions in sales to others and, in a few cases, provided both. The costs and complexities of coronavirus vaccine distribution, also usually paid for by companies, will instead be shouldered by the U.S. Department of Defense.

If our society is bearing the risks and costs for the drug companies, then it should also be able to dictate — not just suggest — that the most incisive clinical studies are performed. In this case, of course, society would also fund the comparative study, including the costs of the vaccines included in it. That would leave no further excuses for not doing this assessment.

Even true differences in vaccine effectiveness, such as a few percentage points gap in immunity rates, will not be apparent when comparing the findings of the various stand-alone vaccine studies. But small actual differences detectable in a comparative study would snowball into enormous differences in benefits. Thousands of deaths will be averted if we identify those vaccines that are more effective, even by a few percentage points.

The potential benefit of finding those differences is an order of magnitude larger when you consider the number of people who could be spared long-term complications from severe Covid-19 infection, who are coming to be known as long haulers. And it is several orders of magnitude greater when you consider the number of students who could safely be attending school rather than staring at computer monitors because we found the most effective path to herd immunity.

Our nation and the world eagerly await the day when vaccines against Covid-19 become available. We need to plan for the day after, when a study compares them head to head — whether the companies that make the vaccines want it or not.

Peter B. Bach is a physician at Memorial Sloan Kettering Cancer Center, where he directs the Drug Pricing Lab.

  • “Pharmaceutical companies almost never voluntarily compare their products against their competitors’.”
    While this is largely true of vaccines, it is often not the case for medicines -except of course in the US. Companies seeking coverage /reimbursement listing in countries that use HTA understand that it is an inherently comparative process, and without head to head data they are unlikely to be able to demonstrate superior effectiveness.
    In Australia, legislation mandates the a new medicine that is more costly than alternatives may not be listed on the national reimbursement formulary unless it provides additional benefit (greater efficacy, reduced toxicity, or both) for at least some patients. (If it can demonstrate that it is no worse than the comparator it may still be listed but not at a higher price) The way to demonstrate superior effectiveness is with one or more head to head RCTs. Indirect comparisons don’t cut it.
    That said, while head to head trials of candidate COVID vaccines would be welcome, and highly desirable in terms of assessing VfM, the need for speed and rapid scale-up is such that this seems like a luxury we cannot afford right now.
    I would be designing some natural experiments, comparing populations receiving different vaccines; then if the results show clinically significant differences in effectiveness, governments should insist on rebates and price reductions.

  • I’m a layman, so sorry if this is dumb….

    Surely what we need now are vaccinations, not vaccines per se? We clearly want to see which is the best product(s), particularly for those most vulnerable. But right here and right now this seems to me to be a second order concern. Picking the bones out of what emerges from OWS (and indeed from other countries too) matters greatly, but it strikes me as something that can wait.

    I hope that in due course better products come along, if anything I’d be surprised if they didn’t. But we are where we are. I am 40 years old and in good health. I’m fully reconciled to the idea that I’m not going to be getting the gold standard product whilst there are unvaccinated highly-vulnerable people. That’s just how it should be.

    So whilst I think we do need to compare, it’s not the priority.

Comments are closed.