Selection bias in occupational sperm studies.

Participation rates in sperm studies are typically 25-50%, and therefore it is a matter of concern whether the men who provide semen samples truly reflect the parent population. The authors analyzed data from three Danish occupational sperm studies and evaluated the relation between age and occurrence of subfertility to participation and provision of semen samples. Age and subfertility were found to be rather strong determinants of participation. Willingness to provide semen samples was greater among men aged <40 years (odds ratio (OR) = 1.8, 95% confidence interval (CI) 1.3-2.6) and among men who had experienced an infertile period (OR = 1.7, 95% CI 1.3-2.1). Furthermore, the effect of infertility was modified by occupational exposure status, thus resulting in a tendency to differential selection and possibly biased risk estimates. In a cross-sectional study, the authors recommend that priority be given to a high participation rate and that data on the basic variables be collected from the entire study population, so that it is possible to make a nonresponder analysis and evaluate bias. Such an analysis can be carried out by means of questions enquiring about infertility, genital disorders, and earlier seminal examinations. Although the longitudinal study design is without many of the limitations of the cross-sectional study, a longitudinal study is usually not feasible.