R Asher Weiss has a teshuva on a similar topic (Shut Minchat Asher, vol. 1, page 126). He answers how one can drink milk from a herd when one knows statistically some of the cows will be tereifot.
He explains that as long as the majority of the cows are kosher, the statistics of what percentage are really tereifot don't matter. Specifically he writes that even if it is true that from a statistical probability it is impossible to say that all cows in a herd are kosher, we can rely on the concept that this herd follows the general statistics of the average number of tereifot in a herd. From a din Torah perspective, every cow has the presumption of being kosher (since statistically more than 50% have been found to be kosher), thus even when bringing many of these cows together, they are all kosher from a Torah perspective.
I see now that Halachipedia mentions a similar argument but see there for other arguments.
Rav Asher Weiss (Minchat Asher Shemot siman 43)
argues that principally we view each cow as kosher because of kol
d’parish m’ruba parish and even when their milk is mixed up the milk
from each cow retains its kosher status despite the overall statistic.
His primary argument is that once we have a halachic principle to
state that the milk is kosher the halacha ignores the physical reality
or statistic. One proof for his argument can be derived from Gemara
Zevachim 73b which implies that once the principle kol d’parish m’ruba
parish is employed even if the items are later mixed up, the remain
kosher. Another proof is the Rama who states that the cheese made from
many unchecked cows who were only considered kosher by merit of a
chazaka remains kosher. This gemara is cited by the Rosh (Chullin
7:37) for a similar point.
I also see R Michoel Zylberman from YU's Kollel wrote an entire article on the topic in the Journal of Halacha bringing different proofs and concluding with the reasoning above. See especially pp. 106 and following.