Using longitudinal data to define the perimenopause by menstrual cycle characteristics.

OBJECTIVES: To determine which aspects of menstrual change best predict time to postmenopause.
METHODS: A total of 250 Australian-born women aged 45-55 years were divided into five menstrual status categories: Group I reported no change in menstrual flow or frequency; Group II reported change in flow; Group III reported change in frequency; Group IV reported change in both frequency and flow; and Group V reported between 3 and 11 months of amenorrhea. Menstrual status groups were compared on baseline data for age, hormone levels, hot flushes and self-rated menopausal status. The proportion of women moving to postmenopause in subsequent years was compared using 4 years of follow-up data.
RESULTS: Women in Group V were older, had lower estradiol and inhibin levels, higher follicle stimulating hormone levels, and were more likely to report hot flushes, and to self-rate themselves as having started the menopausal transition, compared with the women who had menstruated in the last 3 months (Groups I-IV). Groups I and II were similar in age and hormonal status, as were Groups III and IV. The proportion of women who had moved to postmenopausal status in the 4 years after baseline were 12%, 14%, 58%, 53% and 94% for Groups I-V, respectively.
CONCLUSIONS: Amenorrhea is the best predictor of future menopause followed by changes in menstrual frequency. Change in flow only was not predictive of future menopause. A two-stage classification scheme is suggested for defining the perimenopause. 'Early perimenopause' is defined as the self-reporting of changes in menstrual frequency over the last year, and 'late perimenopause' is defined as the self-report of 3-11 months of amenorrhea.