Falling sperm qualityBMJ 1994; 309 doi: https://doi.org/10.1136/bmj.309.6952.476 (Published 13 August 1994) Cite this as: BMJ 1994;309:476
EDITOR, - It has been suggested that environmental factors, possibly acting in fetal and early neonatal life, may be having long term adverse effects on the male reproductive system,1 and that this is leading to an observed deterioration in semen quality2 together with an increase in the incidence of congenital malformations of the male reproductive tract, and in the incidence of testicular cancer. Although this is currently little more than a compelling hypothesis, if true it has very profound implications for those involved in fundamental research as well as those involved in the clinical care of men with reproductive disorders. We thus read with great interest the paper by Peter Bromwich and colleagues,3 together with the associated editorial by Stephen Farrow.4
Bromwich and colleagues make much of the statistical observation, well known to workers in the field of andrology, that measures of human semen quality are seldom normally distributed, and thus correctly point out that the mean is an inappropriate measure of central tendency to use. We entirely agree, although it is interesting that the authors do not in fact test their own observation that a sample of 235 sperm concentrations is log normally distributed; thus their advocacy of the geometric mean (the antilog of the mean of the log data) as the preferred measure of central tendency is open to discussion.
The hypothesis advanced by Sharpe and Skakkebaeck was that factors active in fetal and early neonatal life may adversely affect the subsequent spermatogenic capacity of a given individual.1 If this is true, and if the effect is genuinely becoming more pronounced, the trend in semen quality would be related to the time of the man's birth, rather than the time of any one ejaculation.
To examine this issue, we have looked at data on 3729 semen samples submitted by a large group of semen donors born between 1940 and 1969. These donors were selected only on the basis that they had an initial sperm concentration greater than 20 × 10(super 6)/ml, and all semen analyses were performed in one laboratory, using the same technique throughout.5 When the distribution of the data was examined, it did not differ from a gamma distribution (Kolmogorov-Smirnov d=0.011, P>0.05). We then grouped the samples by year of donor's birth, and plotted the median (interquartile range) sperm concentration for each group (figure). When we compared these groups by using Kruskal-Wallis analysis of variance by ranks, the apparent fall in sperm concentration from a median of 128 in men born in the 1940s to a median of 75 in men born in the late 1960s was significant (H=99.85, P<0.0005). Thus we do not accept that the evidence for a fall in sperm concentrations is unconvincing.