https://astralcodexten.substack.com/p/declining-sperm-count-much-more-than Is Sperm Count Declining?

People say it is.

Levine et al 2017 looks at 185 studies of 42935 men between 1973 and 2011, and concludes that average sperm count declined from 99 million sperm/ml at the beginning of the period to 47 million today.

Levine et al 2022 expands the previous analysis to 223 studies and 57,168 men, including research from the developing world. It finds about the same thing.

Source: Figure 3 here

The "et al" includes Dr. Shanna Swan, a professor of public health who has taken the results public in the ominously-named Count Down: How Our Modern World Is Altering Male and Female Reproductive Development, Threatening Sperm Counts, and Imperiling the Future of the Human Race.

Is Declining Sperm Count Really "Imperiling The Future Of The Human Race"?

Swan's point is that if sperm counts get too low, presumably it will be hard to have babies (though IVF should still work).

How long do we have?

This graph (source) shows pregnancy rate by sperm count per artificial insemination cycle. It seems to plateau around 30 million.

An average ejaculation is 3 ml, so total sperm count is 3x sperm/ml. Since sperm/ml has gone down from 99 million to 47 million, total count has gone down from ~300 million to ~150 million.

150 million is still much more than 30 million, but sperm count seems to have a wide distribution, so it's possible that some of the bottom end of the distribution is being pushed over the line where it has fertility implications.

But Willy Chertman has a long analysis of fertility trends here, and concludes that there's no sign of a biological decline. Either the sperm count distribution isn't wide enough to push a substantial number of people below the 30 million bar, or something else is wrong with the theory.

Levine et al model the sperm decline as linear. If they're right, we have about 10 - 20 more years before the median reaches the plateau's edge where fertility decreases, and about 10 years after that before it reaches zero. Developing countries might have a little longer.

It feels wrong to me to model this linearly, although I can't explain exactly why besides "it means sperm will reach precisely 0 in thirty years, which is surely false". The authors don't seem to be too attached to linearity, saying that "Adding a quadratic or cubic function of year to meta-regression model did not substantially change the association between year and SC or improve the model fit".

Still, the 2022 meta-analysis found that the trend was, if anything, speeding up with time, so it doesn't seem to be obviously sublinear.