Authors' reply
BMJ 1995; 311 doi: https://doi.org/10.1136/bmj.311.7004.570 (Published 26 August 1995) Cite this as: BMJ 1995;311:570
- Niels Keiding,
- Niels E Skakkebaek
- Professor Department of Biostatistics, University of Copenhagen, DK-2200 Copenahgen, Denmark
- Professor University Department of Growth and Reproduction, Section GR-5064, Rigshospitalet, DK-2100 Copenhagen, Denmark
EDITOR,--As Alexander Lerchl points out, figure 1 of our overview indicating that sperm concentrations have decreased during the past 50 years is deficient. During the final preparation for publication, for reasons that we cannot trace, some of the points were omitted. The regression analysis in the paper is unaffected by this: the new regression line (weighted by the number of subjects) had a slope of −0.934x106/ml per year (SE 0.157; P<0.0001), and that line was correctly included in our original figure. A better impression of the regression analysis is provided by the figure in this letter, in which the areas of the circles are proportional to the number of subjects in each publication. There is no reason for Lerchl's scepticism.
Lerchl quotes several criticisms of our paper but omits our detailed and specific responses as well as the subsequently published empirical evidence, which points in the same direction as our paper.
Specifically, Lerchl quotes Brake and Krause, who, on the basis of our data, claimed that sperm concentration had significantly increased since 1970. In fact, Brake and Krause made a mistake in their calculation: the increase they quoted is non-significant (P=0.36). Lerchl quotes Bromwich et al, who offered a speculative, elementary statistical argument with no empirical basis or verification. Lerchl fails to quote our earlier detailed comments on this theoretical exercise.1 2 Lerchl finally quotes the recent report by Olsen et al, who also did not add new empirical evidence: they performed various unsurprising reanalyses of our data, all of which agreed about a significant decline in sperm concentration. We have submitted detailed comments on these reanalyses elsewhere.
Linear regression of mean sperm density reported in 61 publications (represented by circles whose area is proportional to number of subjects in study), each weighted according to number of subjects, 1938-90
Lerchl omits to refer to the additional empirical evidence that has been published. Auger et al (who were originally motivated by serious scepticism about our original report) studied 1351 healthy men volunteering to donate sperm in one clinic in Paris between 1973 and 1992.3 Carefully separating age effects from cohort effects (year of birth), they documented a highly significant decrease in sperm count of 2.1% per year (from 89x106/ml in 1973 to 60x106/ml in 1992) and concomitant decreases in the percentages of mobile and normal spermatozoa. Three additional, shorter reports have been published, also based on data from one clinic and all with similar conclusions.
In a recent international effort the temporal trends in semen quality were viewed in a broader context.4 There have been similar temporal increases in the incidence of testicular cancer and frequently of hypospadias and cryptorchidism, and geographical covariation of several of these symptoms as well as male breast cancer has been documented. In our view it would be irresponsible to disregard this evidence, even if the link to possible determinants is far from definitively established.
Although Lerchl points out a (qualitatively unimportant) deficiency in figure 1 of our paper, we hope that this will not delay a dedicated, wide ranging research effort to clarify these issues.