Single or double semen samples: the dilemma in epidemiological studies on semen quality

Study question: Is it necessary to collect more than one semen sample from each individual in epidemiological studies?

Summary answer: Single semen samples can be used in model construction as long as the model adjusts for major relevant covariates.

What is known already: If a series of semen samples are taken from one individual at different times, significant intra-individual variation may exist.

Study design, size, duration: A population-based cross-sectional study was conducted from 1 December 2000 to 20 November 2002. A total of 1010 volunteers, aged between 20 and 60 years old, were recruited.

Participants/materials, setting, methods: A total of 998 subjects were included for statistical analysis. Of these, 332 men provided single semen samples (group 1) and 666 men provided two samples (group 2). Semen parameters, including semen volume, sperm concentration, total sperm number, progressive motility, vitality, and sperm morphology, were analyzed with standardized methods according to the World Health Organization (WHO) laboratory manual for the examination of human semen and sperm-cervical mucus interaction (3rd edition).

Main results and the role of chance: Rapidly progressive motility of semen from group 1 was lower than that of group 2, but other parameters were not significantly different. The differences in mean values between the double samples were small on all semen parameters. The correlations between the two ejaculates were high, but the agreements were poor except for sperm vitality and normal morphology. Model selection and validation analysis supported the premise that the regression model from the first samples was applicable for the second samples. The large sample size with extensive quality control provides robust parameter estimation and promises good applicability for model selection and validation.

Limitations, reasons for caution: The results are valid for healthy men in China, not for subfertile or infertile men.

Wider implications of the findings: The overall semen quality of a given group of healthy men will not vary significantly in the short term. Single semen samples can be used in model building as long as major relevant covariates are considered. Utilizing multiple semen donations may actually cause the samples to be less representative.

Study funding/competing interests: This research was supported by a grant from Shanghai Science and Technical Committee (02DJ14053) and a grant from Ministry of Science and Technology, China (9902). All authors have no competing interests.