Intraobserver, interobserver variation of sperm critical morphology: Comparison of examiner and computer-assisted analysis

Objective: To examine intraobservational and interobservational variation in sperm critical morphology analysis between trained andrologists and a computer over the range of 0% to 15% normal forms. Design: Retrospective. Setting: Fertility center. Patients: Twenty-four semen specimen slides were read five times in a randomized, blinded fashion by two andrologists and a computerized semen analyzer. Twenty-five samples were analyzed using a different stain and a centrifugation step. Main Outcome Measure: Sperm strict morphology. Results: Mean average intraobserver deviation of normal forms was small (1.3% to 2.7%). Intraobserver deviation did not differ between andrologists, but both differed from the computer. Interobserver variation for percent normal forms was similar for andrologists and computer. Mean average deviations were small over the entire range of normal forms (<3%), there were no differences between examiners, and the variation was independent of the initial percent normal forms. The computer frequently evaluated significantly fewer cells than the andrologists. Updated software increased the number of cells read by the computer requiring more time than the andrologist. Conclusions: Trained andrologists and a computer determine sperm critical morphology similarly over the range of 0% to 15% normal forms. A single evaluation is highly predictive of multiple evaluations. Improvements in the speed of computer assessment are still necessary.