Background: This study aims to determine whether pre or post-processing semen parameters obtained during intrauterine insemination (IUI) predict pregnancy when controlling for confounding effects.

Materials and methods: A prospective cohort study of 2231 semen analyses was conducted at McGill University of IVF center. Any couples who underwent IUI with partner sperm, over a 2.5-year period, were included. Controlled ovarian stimulation was done with Clomiphene Citrate, Letrozole, or Gonadotropins. Statistical analysis was performed using t tests, two types of stepwise logistic regression, and stepwise discriminant analysis. A comparison of pre and post-processing semen parameters was undertaken to determine the probability of pregnancy.

Results: There were significant differences between pregnant and non-pregnant women in post-processing concentration (P=0.043), post-processing total motile sperm count (TMSC) (P=0.049), and post-linearity (P=0.012). However, when variable out-of-the-equation logistic regression or discriminant analysis, which controls for confounding effects between variables, were used, the findings were no longer significant. It was statistically proven that when a variable in the equation logistic regression was employed, post-processing concentration (P=0.005) and post-processing TMSC (P=0.009) remained reliable predictors of pregnancy.

Conclusion: Two of three prediction models suggested that TMSC's relationship with pregnancy is due to confounding factors. One model maintained the validity of the TMSC. While TMSC has always been studied as an important predictor of insemination pregnancies, this finding may be due to confounding effects between semen parameters and therefore requires further investigation as to this relationship.