[lancebeet]

It seems a little strange to me that the people interviewed in this article find this is so remarkable. Only 40,000 people belonged to the group that got the vaccine before 14. If the incidence rate were the same as for the unvaccinated (8.4 per 100k) you would have expected roughly 4 in the group to have gotten cancer. If the incidence rate were the same as for those vaccinated after 14 (3.2 per 100k), you would have expected only ~1 to have gotten cancer. Maybe I'm missing something here?

[dyslexit]

The incidence rate is per year and the study was over an average of 32 years per patient. At that rate you'd expect around 105 incidents instead of 4, although there are other considerations like age to take into account.

[NoZebra120vClip]

The article says that all participants were eligible for screening, but it doesn't say how many were screened or tested.

Also, this cohort would be age 28-36 years currently. The lowest median age for HPV cancer diagnosis is 50. https://www.cdc.gov/cancer/hpv/statistics/age.htm

[bigbillheck]

I believe that's "8.4 per 100k PER YEAR".

[NoZebra120vClip]

The paper uses the term "person-year". So, the intersection of both.

[nerdponx]

> Maybe I'm missing something here?

Statistical literacy is difficult to develop, statistical intuition even moreso, even among well-intentioned and very intelligent people.

[dekhn]

As demonstrated by the original comment we're all replying to (see the other replies).

[logicchains]

That's why papers get away with making big claims that are unsupported by the data.