I’ll say it upfront: yes, I think the paper was a hot potato. Here’s the trail so you can decide for yourself.
Every year, CDC-linked networks publish estimates of how well last season’s flu shot worked. Those numbers usually come from a design called a test-negative case-control study. Across many of those papers, researchers treated early respiratory events differently from later ones — typically the first two weeks after vaccination — on the idea that protection shouldn’t kick in until immunity builds.
Months ago I noticed that special handling creates a classic bias: immortal time. To make the problem visible, I used directed acyclic graphs (DAGs), the causal diagrams epidemiologists have relied on since the 1990s. The short paper — titled, on purpose, a little sharply — “Immortal time bias in test-negative studies of the flu vaccine” — boiled down to three points:
Immortal time is an overlooked bias in these flu-vaccine test-negative studies. Depending on how early events are handled, the causal structure looks like misclassification or selection bias. You can avoid it by counting all events and estimating effectiveness day by day after vaccination as protection builds.
I sent the manuscript to three respected epidemiology journals. Each editor-in-chief bounced it within a week with boilerplate language — no peer review. Why?
Three explanations come to mind: the finding wasn’t important enough; the writing was weak; or the editors decided the bias simply didn’t exist and saw no need for outside reviewers.
Importance isn’t the issue. Flagging a systematic flaw in the annual flu-effectiveness literature matters. Writing quality isn’t a serious explanation either — I’ve published widely in this field, written books, and even served as an associate editor for one of those journals. That leaves the third option: three editorial desks independently treating the analysis as obviously wrong without sending it out. How likely is that unanimous snap judgment?
My own record in DAGs and bias taxonomy doesn’t scream “obviously wrong.” If anything, the analysis is hard to dismiss on methodological grounds. So why the desk rejections?
Here’s the speculation: the paper’s punchline undercuts CDC-associated flu-vaccine estimates. If those analytical choices were mistaken, the published numbers are biased. A published critique could draw press. That is a hot potato. Desk-rejecting avoids the risk that peer reviewers would green-light the piece and force publication.
On a fourth try, I wrote a cover letter stressing novelty and asking for peer review. The wait was longer. The answer was still no — this time with competing reasons: “priority,” space limits, and also that the manuscript didn’t fit the journal’s format closely enough to warrant external review. So which was it — priority, format, or the hot potato?
I stopped shopping journals and posted a preprint. It’s public, but without the peer-reviewed stamp fewer people will see it, and CDC-linked networks will keep reporting estimates built on that design (and other problems besides). Every editor who rejected it, and the authors of those CDC papers, will get the preprint. If someone thinks the analysis fails, they can say so. Silence is its own kind of answer — fine for me, less fine for public health.
Scientific dissent still has room to breathe. Not everywhere, and not without friction — but still.
Brownstone Institute — Eyal Shahar
82.315 SRY