Trends in sperm count in Finnish men

Dear Editor,

Jorgensen et al. recently reported a trend towards lower sperm concentration from 1998 to 2006 among 18–19 year old army draftees from the Turku area in Finland (Jorgensen et al., 2011). This important observation indicates that semen quality may have been higher in the past. Most previous studies on secular trends have been based on samples coming from various populations that need not provide valid comparisons over time (Handelsman, 2001; Fisch, 2008).

Human semen quality is inferior comparing with other mammals (Sharpe, 1994; Joffe, 2010), and it is reasonable to speculate, that human sperm concentrations were higher in the past due to selective evolutionary forces (Czeizel & Rothman, 2002) although such effects may not cause major shifts in sperm counts across few decades (Slama & Leridon, 2002). If so, have modern avoidable exposures during recent decades impaired human spermatogenesis? The Finnish data came from the same population and although participation rates were low (<15%), these young men may not know enough about their fecundity to produce selection bias. Nevertheless, information on the participation rates within each time period is of interest to evaluate differential selection. The semen laboratory took part in a quality-control programme showing systematic bias, but probably not of a type that would generate a trend. Presenting deviations from the quality control standard at each time period would aid evaluation of a drift in laboratory results even though this is not expected.

The observed median sperm concentration among the young men was 60 mill/mL in 1998–99, 54 mill/mL in 2001–03 and 50 mill/mL in 2006, and the corresponding values adjusted for period of sexual abstinence and age were 67, 60 and 48 mill/mL. The difference between the latter was statistically significant. As transformation and adjustment almost doubled the range between highest and lowest median value, it is of interest to know what changed the estimates so much. Are results artefacts of the chosen statistical model? Thus, it would be of interest to compare the crude means, the unadjusted back-transformed means and adjusted back-transformed means to explore effects of transformation and adjustment separately. A model adjusted for some conventional potential confounders such as season, pre- and postnatal smoking and alcohol beverages and analyses of trends in serum concentrations of independent biological markers of spermatogenesis, as Inhibin B, would also be of interest. After all, the fluctuations in crude median sperm concentration values are not large, considering the huge variation of sperm concentration within and between men.

We kindly ask the authors to provide more information and some additional analyses because these data are important, and their interpretation has wide ranging implications. Furthermore, we encourage the authors to make their raw data available for others (Walport & Brest, 2011), especially as the analytical strategy differs between scientific groups.