Fertility prognosis for infertile men: results of follow-up study of semen analysis in infertile men from two different populations evaluated by the Cox regression model.

Using the Cox proportional hazard regression model on one material (group I = 765, 1950 to 1951) we have identified four variables of semen analysis with significant prognostic information about fertility. The four variables were combined into a model for establishing the probability of the individual male to achieve pregnancy as a function of time. This model is tested first on another material (group II = 321, 1977 to 1985). A Goodness-of-fit test indicates excellent agreement between the expectations from the model and the observed number of pregnancies in group II. Second, the two groups are pooled (= 1,086). Then only three variables give significant prognostic information about the time until pregnancy: (1) the man's age at semen analysis (years); (2) the percentage of morphologically normal spermatozoa (ln %); and (3) the degree of motility (good/poor). These three important variables enter into a new and better prognostic model.