Response: A Reanalysis of Sperm Density Data

Male fertility potential in terms of semen quality: a review of the past, a study of the present. Data from men in greater Seattle area reveals no downward trend in semen quality: further evidence that deterioration of semen quality is not geographically uniform. Longitudinal study of semen quality in Wisconsin men over one decade. Wis Med J 91:477479 (1992). 15. Lipshultz Ll. 'The debate continues"-the continuing debate over the possible decline in semen quality. The recent paper by Swan et al. (1) reana-lyzed highly controversial data originally presented by Carlsen et al. (2) as a meta-analysis of sperm counts obtained from the literature and suggested that sperm counts have undergone a worldwide decline approximately over the past 50 years. In the original publication, Carlsen et al. (2) analyzed 61 papers covering five decades of data from multiple countries and found an overall decrease in sperm density. These data were subject to both geographic and temporal bias, as most of the studies in the initial decades of the study were conducted in New York and the majority of the reports were published after 1970. Obvious geographical differences could be the source of bias in the analysis as shown in the reanalysis of the Carlsen data by Fisch et al. (3). Because we have no information regarding potential regional, racial, and/or ethnic variations in normal sperm count that may exist, it is impossible to ignore the potential impact of this type of confounder. If the earlier studies are excluded from the Carlsen interpretation, a second linear regression analysis detects no decline in sperm density (3). In a yet another reanalysis of the data obtained from Carlsen et al. (2) but using different statistical approaches, Olsen et al. (4) found that the data were robust enough to analyze only during the last 20 years, representing 88% of the total number of subjects. They found a lack of the diminishing tendency in all of their statistical models except the for the linear regression model proposed by Carlsen et al. (2. In this latest reanalysis of data, Swan et al. (1) used a multiple regression analysis and stratified the information by geographic region. Although these authors made a very detailed analysis ofmany ofthe possible confounding factors (i.e., length of sexual abstinence , age, methods used to count sperm), it is difficult to rule them out definitively as having an impact on the results. There are …