Do differences between laboratories in semen analysis results matter?

Abstract

OBJECTIVE: Results of semen analyses can vary considerably between laboratories. In this feasibility study, we calculated laboratory-specific factors that correct for these differences and evaluated to what extent the use of these factors improved the value of semen analysis for predicting spontaneous pregnancy.DESIGN: Prospective multicenter cohort study.MATERIALS AND METHODS: 11 Laboratories assessed sperm concentration and morphology of a standardized set of 10 different semen analyses according to World Health Organisation criteria. We computed a z-value for each laboratory from these analyses. Using data from a prospective cohort study of couples presenting for subfertility, we estimated associations with multivariable Cox regression analysis between the results of the male partner's semen analysis and natural conception within one year. We compared model I, which included the original parameters semen volume, and sperm concentration, morphology and motility to model II with z-value adjusted concentration and morphology. To compare the performance of the two models, we calculated the goodness-of-fit (-2loglikelihood) and the area under the receiver operating characteristic curve with 95% confidence intervals and constructed calibration plots.RESULTS: We completed follow-up of 1357 out of 1426 couples (95%), of whom 252 (19%) had an ongoing pregnancy without treatment. The probabilities for an ongoing pregnancy ranged from 4% to 65% for both models. The goodness-of-fit of the first model was slightly better than the second model (-2loglikelihood respectively 3264 and 3267). The AUC for the first model was 0.62 (95% CI 0.58 to 0.66) and for the second model 0.61 (95% CI 0.56 to 0.65). The calibration plots illustrated no difference in performance between the first and the second model.CONCLUSIONS: Z-value based correction of sperm concentration and morphology for differences between laboratories in semen analysis does not substantially improve the predictive capacity of the semen parameters for spontaneous pregnancy. OBJECTIVE: Results of semen analyses can vary considerably between laboratories. In this feasibility study, we calculated laboratory-specific factors that correct for these differences and evaluated to what extent the use of these factors improved the value of semen analysis for predicting spontaneous pregnancy. DESIGN: Prospective multicenter cohort study. MATERIALS AND METHODS: 11 Laboratories assessed sperm concentration and morphology of a standardized set of 10 different semen analyses according to World Health Organisation criteria. We computed a z-value for each laboratory from these analyses. Using data from a prospective cohort study of couples presenting for subfertility, we estimated associations with multivariable Cox regression analysis between the results of the male partner's semen analysis and natural conception within one year. We compared model I, which included the original parameters semen volume, and sperm concentration, morphology and motility to model II with z-value adjusted concentration and morphology. To compare the performance of the two models, we calculated the goodness-of-fit (-2loglikelihood) and the area under the receiver operating characteristic curve with 95% confidence intervals and constructed calibration plots. RESULTS: We completed follow-up of 1357 out of 1426 couples (95%), of whom 252 (19%) had an ongoing pregnancy without treatment. The probabilities for an ongoing pregnancy ranged from 4% to 65% for both models. The goodness-of-fit of the first model was slightly better than the second model (-2loglikelihood respectively 3264 and 3267). The AUC for the first model was 0.62 (95% CI 0.58 to 0.66) and for the second model 0.61 (95% CI 0.56 to 0.65). The calibration plots illustrated no difference in performance between the first and the second model. CONCLUSIONS: Z-value based correction of sperm concentration and morphology for differences between laboratories in semen analysis does not substantially improve the predictive capacity of the semen parameters for spontaneous pregnancy.