Predicting the age at natural menopause in middle-aged women.

Autor:	Hyvärinen M; Gerontology Research Center and Faculty of Sport and Health Sciences, University of Jyväskylä, Jyväskylä, Finland., Karvanen J; Department of Mathematics and Statistics, University of Jyväskylä, Jyväskylä, Finland., Aukee P; Department of Obstetrics and Gynecology, Pelvic Floor Research and Therapy Unit, Central Finland Central Hospital, Jyväskylä, Finland., Tammelin TH; LIKES Research Centre for Physical Activity and Health, Jyväskylä, Finland., Sipilä S; Gerontology Research Center and Faculty of Sport and Health Sciences, University of Jyväskylä, Jyväskylä, Finland., Kujala UM; Faculty of Sport and Health Sciences, University of Jyväskylä, Jyväskylä, Finland., Kovanen V; Faculty of Sport and Health Sciences, University of Jyväskylä, Jyväskylä, Finland., Rantalainen T; Gerontology Research Center and Faculty of Sport and Health Sciences, University of Jyväskylä, Jyväskylä, Finland., Laakkonen EK; Gerontology Research Center and Faculty of Sport and Health Sciences, University of Jyväskylä, Jyväskylä, Finland.
Jazyk:	angličtina
Zdroj:	Menopause (New York, N.Y.) [Menopause] 2021 Apr 12; Vol. 28 (7), pp. 792-799. Date of Electronic Publication: 2021 Apr 12.
DOI:	10.1097/GME.0000000000001774
Abstrakt:	Objective: To predict the age at natural menopause (ANM). Methods: Cox models with time-dependent covariates were utilized for ANM prediction using longitudinal data from 47 to 55-year-old women (n = 279) participating in the Estrogenic Regulation of Muscle Apoptosis study. The ANM was assessed retrospectively for 105 women using bleeding diaries. The predictors were chosen from the set of 32 covariates by using the lasso regression (model 1). Another easy-to-access model (model 2) was created by using a subset of 16 self-reported covariates. The predictive performance was quantified with c-indices and by studying the means and standard deviations of absolute errors (MAE ± SD) between the predicted and observed ANM. Results: Both models included alcohol consumption, vasomotor symptoms, self-reported physical activity, and relationship status as predictors. Model 1 also included estradiol and follicle-stimulating hormone levels as well as SD of menstrual cycle length, while model 2 included smoking, education, and the use of hormonal contraception as additional predictors. The mean c-indices of 0.76 (95% CI 0.71-0.81) for model 1 and 0.70 (95% CI 0.65-0.75) for model 2 indicated good concordance between the predicted and observed values. MAEs of 0.56 ± 0.49 and 0.62 ± 0.54 years respectively for model 1 and 2 were clearly smaller than the MAE for predicted sample mean (1.58 ± 1.02). Conclusions: In addition to sex hormone levels, irregularity of menstrual cycle, and menopausal symptoms, also life habits and socioeconomic factors may provide useful information for ANM prediction. The suggested approach could add value for clinicians' decision making related to the use of contraception and treatments for menopausal symptoms in perimenopausal women. Competing Interests: Financial disclosure/conflicts of interest: J.K. has received funding from Biogen Finland and Tale Ltd. The other authors have nothing to disclose. (Copyright © 2021 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of The North American Menopause Society.)
Databáze:	MEDLINE
Externí odkaz:	Zobrazit plný text záznamu