A Big, Negative Trial

One of the many misinformed complaints of OA advocates is that journals only publish positive findings and squelch negative studies — studies that fail to find differences, effects, or new approaches.

For those of us who actually have worked in the field day-to-day and know better, this has always been a ridiculous claim. In particular, there were years in my career — some consecutive — when reprint and permissions revenues were way off due to the editors receiving and publishing negative trials by the relative boatload.

After all, who wants to pay to distribute copies of trials that prove a drug or procedure is ineffective — that there is no difference, effect, or new approach?

Yesterday, NEJM published an important negative trial, one showing that Paxlovid doesn’t do anything significantly better than a placebo. This is especially impressive because the study comes from Pfizer, the drug’s manufacturer.

This isn’t a rare occurrence, but I wanted to mention it because it’s a perfect example of how science remains a priority for some, and of how journal publication criticisms are often misinformed and misleading.

Negative trials are important, and they make their ways into the literature when they’re good enough to warrant publication.

How do you like them apples?

Addendum

After writing today’s post about the negative trial of Paxlovid, I took a trip to the gym to get my blood pumping — and realized there are a few more points worth making:

1. There is no clear indication that this was submitted as a negative trial. In fact, there is a readily apparent path to interpreting it as a positive trial using simple statistical techniques most people — especially the lay public — wouldn’t object to on their face. But it has been published as a negative trial, perhaps because the editors, statisticians, and reviewers at NEJM discovered the numbers actually supported that outcome more than a positive interpretation.
   1. One of my favorite moments during my years of sitting in NEJM editorial board deliberations was when the editors and statisticians discussed an important paper making a positive claim, agreed the numbers didn’t add up, and insisted that the authors introduce the word “not” into their conclusion that “x is associated with superior outcomes” or some such construction, so that the published paper read, “x is not associated with superior outcomes.”
2. If this were posted as a preprint per Gates and cOAlition S’ whimsical and reckless policies, would it have been a negative trial? Or would we have seen a trial supporting wider prescription of Paxlovid based on data tilted ever so slightly, almost imperceptibly, in that direction? Would we see a hardening of beliefs, policies, and reimbursement practices benefiting Pfizer? We’ll never know, but one can guess.

I wanted to add these thoughts as an addendum to the earlier post, and to remind myself and others that you think better when your heart is pumping a little faster — so, after some exercise, a hot shower, or an exciting event.

Thanks for reading.