Skip navigation

Tight underpants

Sperm counts in the Western world are falling. Tight underpants? Oestrogens in water? Hormones in food? Desk-bound males? Stress? The 'truth' of this observation has now achieved almost universal acceptance, and the issue has been transformed more into what is the cause and what can we do about it? We've even seen 'projections' that by 2050, or 2150, or whatever, if sperm counts keep falling at the rate they are now, human reproduction will be entirely by cloning.


There's a new analysis which questions this 'truth' [ 1 ]. It revisits the data in an original meta-analysis [ 2 ] which collected 61 studies on sperm counts published between 1938 and 1990. Almost half of these studies (29/61) studied fewer than 50 men. The smallest number was nine and the largest 4435 men.

Originally a linear regression of the sperm count against year of the study showed a significant negative correlation. The new analysis [ 1 ] says that all the curve fitting done to date (several types have been tried) is wrong, and that if a flexible smoothing model is used then it shows that there may be a fall in the USA, but not in other countries.

Hang on a minute

Perplexed by all this, and getting more interested by the minute, Bandolier revisited the original data. Firstly, in the USA 80% of the total of 8416 men were studied after 1970. For the rest of the world it was 98%. So very little information exists before 1970. When we look at weighted mean values by decade or two (Table), all the data between 1970 and 1990 looks solid and reproducible, with no obvious changes going on. Only earlier information looks as if sperm counts were higher, but then in few men.
Sperm counts in USA and other countries from 1930 to 1990
  United States Rest of the World
Period Number of men Weighted mean sperm count (million/mL) Number of men Weighted mean sperm count (million/mL)
1930-1950 496 119 100 95
1951-1970 1184 107 0 no data
1971-1980 1868 72 427 73
1981-1990 4868 67 6004 77

In the original meta-analysis considerable thought is given to possible sources of bias or analytical change to try to find possible origins for the change. They found none. Yet one sentence says that the coefficient of variation for determination of sperm density was 25 in an external quality check between ten German laboratories.

Let's see what that means with a sperm count of, say, 70 million/mL. A coefficient of variation of 25% says that one standard deviation is 25% of 70, or about 18 million/mL. That means that 95% of results from these ten laboratories would be between plus two and minus two standard deviations, or between 34 and 106 million/mL. Yes, the highest and the lowest would vary by just about three-fold.

Close but no cigar!

Does the evidence we have constitute something to worry about? Is the story about falling sperm counts true? Readers can make their own minds up, but Bandolier isn't going to lose sleep over this.


  1. S Becker, K Birhane. A meta-analysis of 61 sperm count studies revisited. Fertility and Sterility 1997 67: 1103-8.
  2. E Carlsen, A Giwercman, N Keiding, NE Skakkebaek. Evidence for decreasing quality of semen during past 50 years. British Medical Journal 1992 305: 609-13.

previous or next story in this issue