Swan, SH, C Brazil, EZ Brobnis, F Liu, RL Kruse, M Hatch, JB Redmon, C Wang, JW Overstreet, and the Study for Future Families Research Group. 2003. Geographic differences in semen quality of fertile US males. Environmental Health Perspectives 111:414-420.

Press coverage
New York Times
Kansas City Star

Swan et al. present conclusive proof of significant differences in sperm count among geographic areas within the United States. Previous studies had suggested geographic variation, but problems in comparing different studies and measurement techniques prevented firm conclusions.

In this new study, Swan et al. developed and applied a rigorous protocol that was shared by labs in four different regions of the country. Their work reveals that men from rural areas of Missouri have lower sperm counts than urban dwellers in New York, Minnesota and Los Angeles.

Why these differences exist or over what time span they developed remains unknown.

What did they do? Swan et al. obtained sperm samples from the husbands of pregnant women coming to a university hospital for prenatal care in the four study areas: New York NY, Minneapolis MN, Los Angeles CA and Columbia, MO. Only husbands at least 18 yrs old were selected, and only if the pregnancy was not medically assisted. Several other criteria were also used consistently at each participating university hospital. The care taken to choose the sample was a key methodological breakthrough for this work because most studies of sperm counts have been based upon samples that were unlikely to be representative of the local population (one notable exception; and see comment below).

Most participants contributed two sperm samples, separated by approximately 3 weeks. Subjects were asked to abstain from ejaculating for 2-5 days prior to providing a semen sample. The time of the last ejaculation prior to providing the sample was obtained by interview as were data on a variety of other factors known to affect sperm count. These variables were all included in the statistical analysis. Men with abstinence times of less than 2 hrs or greater than 10 days were excluded.

The samples were examined at the four university hospitals using procedures that were carefully developed to minimize any measurement artifacts; technicians conducting the measurements were trained and tested for proficiency. They measured sperm density, volume and motility, and also compared sperm morphologies. Extensive statistical analysis and adjustment were used to correct for differences among the centers in confounding variables, such as age, race, smoking, history of fever and sexual-transmitted diseases, genital problems, etc.

What did they find? Semen samples came from 493 men, of whom 410 provided two samples (on average 24 days apart). The average abstinence time overall was 78 hours; each of the four study sites averaged within 6 hours of that overall average.

Sperm concentration, percent motile sperm and total motile count were lower in Missouri than at all other centers (see table below).

Swan et al. measured sperm density using two different techniques, both of which reported the lowest average in Missouri. The table below presents the averages of their measurements using a standardized technique called the µ-Cell (a disposable counting chamber).

In their extensive statistical analyses, adjusting the measurements to correct for the effects of the confounding variables recorded in interviews by Swan et al. did not eliminate the large differences between Columbia MO and the other sites, even though a number of these variables, by themselves, were important.

For example" "Smoking more than ten cigarettes per day was associated with decreased semen volume, but had little effect on concentration, motility or morphology. Fever within the prior three months significantly decreased sperm concentration and motility, but not morphology or semen volume. The percent morphologically normal sperm was reduced among men who reported a history of a STD" [sexually-transmitted disease].

What does it mean? Swan et al. observed striking differences between Missouri men and those from the other study locations. Why? At present, they do not know. The most conspicuous difference between the areas is Columbia, Missouri's, close proximity to industrial agriculture and its pesticides. Yet the data that Swan et al. present in the study do not allow any conclusions to be reached about whether this obvious difference is involved in causing the differences between the regions. They conclude with the observation:

There is no reason to propose that large genetic differences in the population makeup of Missouri vs. the other sites account for the observed geographic patterns... the US population is too mobile. If that is true, then a clear implication of Swan et al.'s data is that sperm count in Missouri has declined, although over what time scale this may have occurred is unknown.

By limiting the study subjects to fertile husbands only, Swan et al. may have reduced the real differences between Missouri and other men. Their selection criteria bias the sample against men with very low sperm counts, particularly those so low they are incapable of fertilization. Given the observed differences between Missouri and other regions, men with extremely low sperm counts are more likely in Missouri than elsewhere. Thus it is possible that the selection criteria led to inclusion within the study of men in Missouri with higher sperm counts than would be obtained were the sample truly random.