Timing of menopause and patterns of menstrual bleeding.

Abstract

Age at menopause is associated with a variety of health outcomes. Menstrual histories, both as markers of physiologic function and through their potential association with age at menopause, have also been investigated for their links to health outcomes. This study used data from a cohort of women from the United States who provided prospectively recorded data on their menstrual cycles for many years. Dr. Alan Treloar (University of Minnesota) originally recruited the women in the 1930s; the authors used data reported by these women from 1930 through 1977. They identified nuanced characteristics of menstrual histories that were strongly predictive of the onset of menopause, focusing on measures of central tendency (the mean), variability (standard deviation), and serial irregularity (approximate entropy), the last of which quantifies a continuum that ranges from totally ordered to completely random. They controlled for other characteristics known to affect age at menopause, including use of exogenous hormones, number of births, and extent of breastfeeding. Although cycle length and variability increased with the approach of menopause, the authors found that serial irregularity decreased and was a strong predictor of its onset. This finding constitutes an important piece of information not attainable with conventional statistical summaries of menstrual histories.