The statistical foundation of the reference population for semen analysis included in the sixth edition of the WHO manual: a critical reappraisal of the evidence

In the most recent version of the 'WHO Laboratory Manual For The Examination And Processing Of Human Semen', the updated target population used to infer reference values included 3589 fertile subjects, representative of 12 countries and 5 continents, and 10 studies. We have critically evaluated the newly proposed distribution of semen examination results using an approach borrowed from clinical chemistry laboratories and based on the recommendations of the International Federation of Clinical Chemistry for estimation of reference intervals. Surprisingly, most prerequisites to produce common reference intervals through multicentric data were not met. Moreover, when we assessed with the bootstrap method the descriptive reference values obtained from raw data of the 10 individual studies for sperm concentration, sperm number, motility and normal forms, we found that none of the populations was completely correctly described by the reference centiles. We concluded that aggregated data used to build the reference distribution cannot be considered to originate from the same population, and this can result from real differences among individuals or different methodological approaches used in the various studies. Transferability conditions across studies did not seem to have been met. Our findings strengthen the relevance of concerns regarding the use of reference populations in the World Health Organization manual to discriminate between fertile and infertile men.