Objective: The variability in ultrasound-based antral follicle counts sized 2-10 mm after allowing for age-related decline is considerable. This may represent differences in actual reproductive age among women. This hypothesis was tested by cohort comparison for distribution of age at occurrence of reproductive events.
Design: A model with a nonlinear mean decline with age was fitted to antral follicle counts (AFC) obtained in 163 regularly cycling fertile volunteers. Ages at last child birth and menopause were predicted from the individual AFC by using thresholds to represent these events and the model for decline with age. Distributions of the observed ages at last childbirth (proxy variable for loss of natural fertility) and ages at menopause were obtained from the BALSAC demographic database and the Prospect-EPIC study, respectively. The observed distributions were compared with the predicted distributions by using visual comparison and quantile-quantile plots. Predictions of age at last child and age at menopause were done using percentiles of the modeled AFC distribution for given age, and corresponding percentiles of the predicted distributions of age at these reproductive events, with predictions following from the position of a woman's AFC relative to these percentiles.
Results: The predicted distributions of age at last child and age at menopause showed good agreement with the observed distributions in the BALSAC and EPIC cohort. Compared with age alone, antral follicle counts gave some additional information for individual prediction of age at last child and menopause.
Conclusions: The link between declining antral follicle counts and reproductively significant events like loss of natural fertility and menopause is strengthened by the high degree of similarity among the predicted and observed age distributions. Predictive usefulness of this relationship in a clinical setting may be more marginal, except in the case of women who have low AFCs for their age.