13 September 2021
Summary
Hyvärinen et al. [1] carried out a study with the aim of predicting the age of natural menopause (ANM) in middle aged women. They used Cox models with dependent covariates, selecting predictors with lasso regression. For this research, authors analyzed the NAMS longitudinal study "Study of estrogen regulation of muscle apoptosis" consisting of 6,878 women initially screened of which 1,393 were finally surveyed and divided into pre- and perimenopausal. Of these, data of 279 participants (45 to 55 years of age) and that of 105 women that completed bleeding diaries were analyzed. Results were quantified with c-indices, mean and standard deviations of absolute errors. Regression models included alcohol consumption, vasomotor symptoms, self-reported physical activity, and relationship status as predictors, as well as, estradiol and follicle-stimulating hormone levels, menstrual cycle length, smoking, education, and the use of hormonal contraception. After analysis of the models, the authors conclude that in addition to the levels of sex hormones, rhythm of the menstrual cycle and menopausal symptoms, lifestyle habits and socioeconomic factors can provide useful information for the prediction of ANM.
