FIG. 3. This bar diagram of ovulation disturbances cross-sectionally shows the percentages of cycles in women of different gynaecological ages that are anovulatory (open bars), have SLPs (short LP, grey bars) and are normally ovulatory (normal, black bars). Note that anovulatory cycles increase in women of gynaecological ages 35—40 but that SLPs increase only slightly. Adapted to represent data by Vollman (1977).
of these cross-sectional studies are the necessary clinical or prospective menstrual cycle data reported that would allow a diagnosis of perimenopause or assessment of the phase of participants.
There are even fewer prospective studies of ovulation and its disturbances during perimenopause. The majority of those that do exist poorly describe recruitment criteria and the participants' physiological characteristics and clinical experiences. Most well designed longitudinal studies report menstrual cycle data without documentation of ovulation (Brambilla et al 1994, Kaufert et al 1987, Treloar 1981). A number of other important longitudinal studies collecting intermittent hormonal data have also not documented ovulation (Guthrie et al 1998, Longcope et al 1986). Even if ovulatory status is determined, luteal phase lengths are not assessed systematically by any longitudinal data sets except those of Vollman (1977). Others have only documented anovulatory but not SLP cycles, because weekly urine sampling does not allow luteal length quantification (Metcalf 1979, Brown 1985).
The available prospective data consistently show that ovulation disturbances increase in perimenopause. Vollman (1977) diagrams the final 100 cycles in one woman who experienced 16% of cycles as anovulatory. Vollman also noted that within one woman the basal temperatures decreased (in both the follicular and luteal phases) in cycles during the 3rd and 4th decades from menopause compared with the 2nd and 1st (Vollman 1977). Temperatures were lower by — 0.1 °C in the follicular and by—0.15 °C in the luteal phase, suggesting an added effect of lower progesterone production in addition to the effect of an apparently lower metabolic rate (documented earlier, Collett 1949).
Although detailed cycle data are graphed for three or four cycles each from only four women, Shideler shows the variability of both flow and ovulation (Shideler et al 1989). Figure 3 shows that each woman in phase C of perimenopause experienced variations in both luteal phase and in cycle lengths and that each experienced ovulation disturbances. These women, in contrast to many reported in the above cross-sectional studies, were perimenopausal by cycle variability and/or vasomotor symptoms (VMS) (Shideler et al 1989).
A much longer set of weekly urinary hormonal data are available from each of two women over the final 6—7 years of the perimenopausal transition (Brown 1985), and from one woman over 108 consecutive weeks (Metcalf & Donald 1979). The two women studied by Brown recorded 15 and 23 episodes of flow and both had several months of no flow at the end of monitoring. Using 5 as the PdG cut-off for ovulatory cycles, 8 of 15 and 10 of 23 cycles — only 48.4% — were ovulatory (Brown 1985). The longitudinal data in one woman, however, appeared to be earlier in the perimenopausal transition and showed that 86% of cycles were ovulatory (18 of 21 flow episodes were ovulatory by PdG) (Metcalf & Donald 1979).
In a larger and more systematic study, and the only one that had as its primary focus as the prospective study of ovulation in perimenopause, Metcalf systematically documented weekly urinary PdG over three cycles in 139 women over 40 years of age (Metcalf 1979). 93% of the 86 participants in whom all of the documented cycles were of normal cycle length (21—35 days) were ovulatory (although some probably were SLP cycles) as were 95% of the cycles in the 81 women who reported no recent cycle changes. However, if cycles were irregular either by history (in 58 women) or by prospective documentation (in 53 women), the percentage of the three cycles that were consistently ovulatory decreased to 39.7% and 34%, respectively (Metcalf 1979). These latter women who were perimenopausal by epidemiological criteria based on irregular cycles had an average of 36.9% of consistently ovulatory cycles (Metcalf 1979). Despite this, PdG levels did not differ with age and 10.3% of women who had gone three months without flow subsequently ovulated (Metcalf 1979).
The final prospective study documented three women aged 37—47 for whom one cycle of daily hormonal data were available approximately 10 years previously and graphically displayed two cycles of hormonal data 11 years apart from one woman (Welt et al 1999). Mean progesterone levels were lower in women studied 10 years later (4.9 versus 11.7 ng/ml) in addition to lower luteal phase inhibin A levels and borderline lower follicular phase inhibin B levels. No significant prospective differences in LH, FSH or oestradiol were documented (Welt et al 1999). From graphed data, luteal phase lengths shortened in one woman followed over 11 years (10 and 11 versus 12 and 13 days from the LH peak) (Welt et al 1999).
A large prospective epidemiological study from Malmo, Sweden, providing the only population-based data of ovulation available, studied over 150 women for 12 years with serum hormone levels measured at 3—6 monthly intervals (Rannevik et al 1995). This study followed women through menopause and for several years following it. In women who were 72—61 months before their final flow, 62.2% of cycles were ovulatory with a mean progesterone level of 27.3 nmol/l. However, women in the 0—6 months before the final £ow experienced ovulatory cycles only 4.8% of the time although the average progesterone level was unchanged at 22.4 nmol/l (Rannevik et al 1995). These data were presented as a summary in the text but no primary progesterone data were provided.
Thus prospective data consistently document ovulation disturbances as a part of perimenopause and reproductive ageing in women. In women who were symptomatic with new cycle irregularity or vasomotor symptoms the incidence of ovulation disturbances was increased. However, the literature contains information about prospective variations in ovulation in relatively few women and includes no primary data from epidemiological samples. These prospective results (Fig. 4) suggest that ovulation disturbances do not generally increase until phase C of perimenopause when cycle intervals start becoming variable.
Was this article helpful?