Secular declines in age at menarche have been well documented and a secular increase in age at menopause has been suggested by some studies. Change in age at menarche or menopause are of scientific interest because their timing is related to women's fertility, risk of chronic diseases, and mortality. An equally important, but rarely addressed question is whether secular change has occurred in menstrual function during the prime reproductive years or the menopausal transition. Menstrual characteristics of adult women are important markers of ovarian function and have been associated with fertility, age at menopause, and women's long term risk of chronic disease. Given the secular change in risk factors that influence menstrual function, including body size, smoking, parity and hormone use, evaluation of their impact on menstrual function is warranted. The historic TREMIN dataset presents a unique opportunity to address the question of secular change in menstrual characteristics of US women. It is the only dataset in existence worldwide with prospectively recorded menstrual cycle information across two distinct generations of women sampled from a comparable referent population and linked to information on reproductive, anthropometric and lifestyle characteristics. Students attending the University of Minnesota were enrolled in 1936-39 (Cohort 1) and in 1961-64 (Cohort II). We propose to analyze the prospectively recorded menstrual calendar and health record data from the 1151 Cohort I women and 905 Cohort II women who participated for 5 to 35 years. Our goal is to evaluate whether secular change has occurred in menstrual patterns of adult women or in the timing of the menopausal transition and, if so, whether these differences are explained by secular differences in parity, body mass index, smoking, or hormone use. We will estimate menstrual characteristics across the lifespan using linear mixed models and marginal models fit with generalized estimating equations to separately model mean, variance and correlation of menstrual cycle length by age. We will estimate the effect of cohort on age at onset of the menopausal transition and on duration of the transition using Cox proportional hazards and the varying coefficient Cox model along with a change-point model for mean and variance. We propose multiple imputation techniques to account for intermittent missing data and new methods for addressing the problem of hormone use masking the timing of reproductive aging. Analyses of these truly unique data will make a substantial contribution to our understanding of the nature of secular change in menstrual characteristics throughout reproductive life. Understanding of the extent to which menstrual function during the prime of reproductive life is influenced by secular change in population risk factors would allow us to predict the potential impact of such changes on fertility, menstrual morbidity and reproductive aging. Ultimately, surveillance of menstrual characteristics young women might facilitate prediction of changes in age-related fertility dynamics, timing of the menopausal transition and chronic disease risk associated with reproductive aging. This project takes advantage of the historic TREMIN dataset -- the only dataset in existence worldwide with prospectively recorded menstrual cycle information across two distinct generations of women sampled from a comparable referent population -- to assess whether secular change has occurred in the menstrual experience of young adult and midlife women. Menstrual characteristics are important markers of ovarian function and are associated with fertility, age at menopause, and women's long term risk of chronic disease. Understanding of the extent to women's menstrual experience is influenced by secular change in population risk factors such as body size, smoking, parity and hormone use will facilitate prediction of the potential impact of these population changes on fertility, menstrual morbidity and reproductive aging. ? ? ?
| Huang, Xiaobi; Elliott, Michael R; Harlow, Siobán D (2014) Modeling Menstrual Cycle Length and Variability at the Approach of Menopause Using Hierarchical Change Point Models. J R Stat Soc Ser C Appl Stat 63:445-466 |
| Huang, Xiaobi; Harlow, Siobán D; Elliott, Michael R (2012) Distinguishing 6 population subgroups by timing and characteristics of the menopausal transition. Am J Epidemiol 175:74-83 |
| Wang, Chia-Ning; Little, Roderick; Nan, Bin et al. (2011) A hot-deck multiple imputation procedure for gaps in longitudinal recurrent event histories. Biometrics 67:1573-82 |
