Comparison of normal sperm morphology outcomes from two different computer-assisted semen analysis systems.

By eliminating the human evaluation variable, it was possible to carry out an investigation into the 'true' association between normal sperm morphology outcomes assessed according to the World Health Organization guidelines and strict criteria. Two computer-assisted semen analysis systems were used, IVOS and Mika, to evaluate Diff-Quik and Papanicolaou stained slides. As expected, the mean normal sperm morphology outcomes for the World Health Organization classification evaluations were markedly higher for both the Diff-Quik (mean difference = 40.13%) and the Papanicolaou (mean difference = 32.55%) stained slides. The association between the outcomes were low for the Diff-Quik stained slides (r = 0.379) and poor for the Papanicolaou stained slides (r = 0.110). While the association achieved with the computer-assisted semen analysis systems using Diff-Quik stained slides was comparable to the association between the manual evaluations (r = 0.386), the manual evaluation of Papanicolaou stained slides produced a relatively good association (r = 0.690). Although the numbers were small, the results show the probability of poor class correlations. Approximately 40% of outcomes were incorrectly classed at a 14% (strict criteria) and 50% (World Health Organization guidelines) cut-off point for both staining methods. This study confirms the fundamental differences between the two classification systems. The results also indicate that, of the two stains used, Diff-Quik should be the preferred staining method for computer-assisted sperm morphology evaluations.