Reducing Inter-Laboratory Differences between Semen Analyses Using Z Score and Regression Transformations

Abstract

Background Standardization of the semen analysis may improve reproducibility. We assessed variability between laboratories in semen analyses and evaluated whether a transformation using Z scores and regression statistics was able to reduce this variability. Materials and Methods We performed a retrospective cohort study. We calculated between-laboratory coefficients of variation (CVB) for sperm concentration and for morphology. Subsequently, we standardized the semen analysis results by calculating laboratory specific Z scores, and by using regression. We used analysis of variance for four semen parameters to assess systematic differences between laboratories before and after the transformations, both in the circulation samples and in the samples obtained in the prospective cohort study in the Netherlands between January 2002 and February 2004. Results The mean CVBwas 7% for sperm concentration (range 3 to 13%) and 32% for sperm morphology (range 18 to 51%). The differences between the laboratories were statistically significant for all semen parameters (all P < 0.001). Standardization using Z scores did not reduce the differences in semen analysis results between the laboratories (all P < 0.001). Conclusion There exists large between-laboratory variability for sperm morphology and small, but statistically significant, between-laboratory variation for sperm concentration. Standardization using Z scores does not eliminate between-laboratory variability.

Keywords

• differences
• semen analysis
• regression
• standardization