Trends in global semen parameter values

Allegations for a worldwide decline in semen parameter values have not withstood scientific scrutiny. Methodological flaws in an influential 1992 paper are summarized here, and studies that have been published since 1992 are reviewed. Of the 35 major studies of time trends in semen quality reviewed here, eight (a total of 18 109 men) suggest a decline in semen quality; 21 (112 386 men) show either no change or an increase in semen quality; and six (26 007 men) show ambiguous or conflicting results. The cause (or causes) of the geographical and temporal variations in semen parameter values reported by these diverse studies deserve further investigation.

This paper critically examines two related scientific assertions: that there have been worldwide declines in semen quality in recent decades and that these declines are related to environmental exposure to minute levels of ‘endocrine disruptors' (chemicals that exhibit some degree of estrogen-like activity in the body). The data supporting these assertions will be demonstrated to be weak, at best. Reported declines in semen parameter values are likely to be either highly local phenomena with an unknown etiology or the result of methodological errors arising from attempts to observe highly variable physical attributes (semen characteristics) with relatively low-resolution tools (retrospective analysis of nonrandomized study populations).

Methodological challenges

For many reasons, semen may be the most poorly understood bodily fluid, in terms of the distribution of its normal values in the general population. One problem is the relative difficulty of obtaining human semen for scientific analysis. The fact that semen is most readily obtained by masturbation poses significant logistical barriers to objective, randomized, longitudinal studies of semen parameters in community-dwelling men. For example, participation rates in the few studies that have attempted to assess semen quality in non-infertile men have typically been <20%).^1,2 Although it is possible, in principle, to conduct large-scale population-based trials, the procedural issues involved are challenging, which is likely the reason that such trials have not yet been conducted. Instead, research to date on semen quality has relied on populations of men who have provided semen samples for sperm donation, infertility evaluation, prevasectomy evaluation, or for other specific reasons. Each of these populations presents a possible selection bias and none represent a random sample of the population at large. For example, semen donors may have been screened for problems known to affect fertility, or they may have been selected precisely because a prior semen analysis indicated robust fertility. Male donors in cases of in vitro fertilization or other attempts to overcome infertility issues, on the other hand, are more likely to have low fertility, regardless of the fertility status of their partner. It is, therefore, fundamentally difficult at present to determine ‘normal' semen parameters for community-dwelling populations of men.

In addition, semen attributes such as sperm count, semen volume and sperm morphology vary widely between individuals as well as within individuals over time. Longer periods of time since the previous ejaculation (abstinence time), for example, are associated with higher sperm counts, higher semen volumes and a higher percentage of sperm displaying abnormal morphology. Other sources of variability include: scrotal temperature,³ season of the year,⁴ smoking status,⁵ marijuana use⁶ and geographic region.⁷ Although some studies of semen parameter values have attempted to control for some of these variables, many have not, which seriously compromises the conclusions that can be drawn from such studies.

Critical analysis of an influential paper

In 1992, Carlsen et al.⁸ published a metaanalysis of 61 previous studies of semen parameters and reported a nearly 50% drop in sperm concentrations, from 113×10⁶ per ml in 1940 to only 66×10⁶ sperm per ml in 1990, and raised the question of whether this ‘decline' might be due to exposure to compounds with estrogen-like activity. Although the paper generated a great deal of media attention, it has been repeatedly criticized in the scientific community for its many methodological flaws.^{9,10,11,12,13} These flaws include: high cross-study variability in the methods and protocols used for sperm collection and measurement; lack of control for period of abstinence, cigarette smoking or recreational drug use; failure to include some studies reporting no decline in semen parameters; and failure to account for geographic variation between studies.

The pronounced geographic variation in semen quality in particular, is a source of serious error. All of the studies included in the meta-analysis from before 1970 were from the United States, and 80% of these were from New York State, where sperm counts (then and now) are higher than average. After 1970, only three studies were from the United States, and many were from third-world countries where sperm counts were lower than average. If the Carlsen data are reanalyzed to account for this geographic variation, no decline in sperm counts is found (Figure 1).⁷

Open in a separate window

Figure 1

Reanalysis of data from Carlsen et al.⁸ showing no decline in sperm concentrations (black regression line) when data from New York are excluded. Bubble size corresponds to number of men in study.⁹

Another potential weakness of the Carlsen study involves the use of an inappropriate linear regression model in the statistical analysis.¹⁴ Because the data distribution was highly nonuniform, quadratic or spline regression models are more appropriate analytical tools. When these tools are applied to the Carlsen data, mean sperm concentrations have actually increased since 1940.¹²

In the two decades since publication of Carlsen's paper, at least 35 major studies of time trends in semen parameters have been published. Eight (on a total of 18 109 men) suggest a decline in semen parameter values (Table 1); 21 (on a total of 112 386 men) show either no change or an increase in semen quality (Table 2); and six (on a total of 26 007 men) show ambiguous or conflicting results (Table 3). (Studies that reanalyzed existing data or that were critiques in general ways of some of the methodological issues involved in the debate over alleged changes in semen parameters were not included in this review.) As with previous studies, all of these investigations rely on populations of men who are not necessarily representative of the general male population.

The evidence provided by these studies refutes the simplistic notion of a worldwide decline in semen parameters, though, clearly, they also demonstrate that semen parameter values vary dramatically both geographically and temporally. These variations may arise from numerous causative factors including: differences in lab techniques and analysis between regions; differences in sexual behavior that alter mean abstinence times between regions; genetic variations between populations; variation in lifestyle factors such as obesity or recreational drug use; or variations in in utero exposure to mutagenic compounds or environmental pollutants. The wide geographic variations in semen parameter values cannot yet be adequately explained; however, they must be adequately controlled for in any metaanalyses or other attempts to draw broad conclusions about worldwide trends.

To date, however, the data supporting a role for ‘endocrine disruptors' in the alleged ‘decline' in semen parameters is weak.¹⁵ Some, but not all, studies, for example, have found no association between variations in semen quality and location in rural vs. urban areas, or between areas with known high levels of air pollution and those with less pollution.¹⁶ Moreover, a potential causal relationship between semen quality and ‘endocrine disruptors' cannot be investigated by studies of semen parameters alone. A recent review of epidemiological studies of changes in semen parameter values and exposure to endocrine disrupters concluded that convincing human evidence that such exposure has an impact on male fertility is still lacking.¹⁷

It is worth nothing that, in contrast to the wide variations in results of studies of semen parameter values, studies of temporal trends in testosterone levels have been more uniform. Data from randomized, adequately sized populations of community-dwelling men show a clear and consistent decline in mean testosterone levels in recent decades.¹⁸ Whether such declines are related to the declines in semen quality reported in a minority of studies reviewed here remains to be scientifically explored.

Conclusion

The allegations for a worldwide decline in semen parameter values presented by Carlsen et al.⁸ in 1992 have not withstood scientific scrutiny. This paper, and others, have detailed the methodological flaws in the Carlsen paper that warrant its exclusion from future reviews of the data pertaining to variations in semen quality over time. In the two decades since publication of Carlsen's paper, at least 35 major studies of time trends in semen parameters have been published. Eight (on a total of 18 109 men) suggest a decline in semen parameters; 21 (on 112 386 men) show either no change or an increase in semen parameters; and six (on 26 007 men) show ambiguous or conflicting results. The cause (or causes) of the geographical and temporal variations in semen parameters reported by these diverse studies deserve further investigation.

Notes

The authors declare no competing financial interests in relation to this paper.

References