MP46-01\u2003THE INITIAL MALE INFERTILITY EVALUATION AND PREGNANCY– WHAT REALLY MATTERS? A STUDY FROM THE UTAH POPULATION DATABASE USING A PREDICTIVE ANALYSIS MODEL

Abstract

INTRODUCTION AND OBJECTIVES: Despite the semen analysis (SA) being the cornerstone of the male infertility workup, there are significant limitations. Large intra-individual variability exists and the ability to predict pregnancy is in question. We sought to determine the most predictive male factors for subsequent pregnancy at five years based on the initial male infertility evaluation. METHODS: We utilized the Subfertility Health and Assisted Reproduction (SHARE) study that was then linked to the Utah Population Database (UPDB). We analyzed only men at their first SA and those seen at the University of Utah after 1998. We excluded azoospermic men and men with missing SA data. Cox regression models with elastic net regularization were used to select predictive variables among a group of highly correlated semen parameters, which cannot be accomplished with traditional regression methods. Results were used to develop a nomogram that can be used to predict the likelihood of conception within 5 years of semen analysis. RESULTS: Our final cohort consisted of 6,861 men, with a median age of 32 (IQR 28-34). The average number of previous children was 0.4 (SD 0.8). 29% of the cohort was a former or current smoker, and 42% of the men were overweight (BMI 25-30) and 16% were obese (BMI>30). The median income was 49,807 USD (IQR 41,901- 56,937). The strongest predictive factors were male age, number of previous children, and median income. Of the SA parameters, total motile count was the most predictive of conception. Never smoking was a stronger predictor than any one individual semen parameter. CONCLUSIONS: By utilizing the long-term follow up in the UPDB, we developed a unique predictive model and nomogram for predicting pregnancy at 5 years based on initial male infertility evaluation. The main drivers of pregnancy at 5 years were age, median income, and previous children. Of the semen parameters included, total motile count contributed the greatest to predicting conception at 5 years. This study is further evidence that questions the clinical utility and predictive value of individual semen parameter Figure. No caption available. Source of Funding: none