https://orcid.org/0000-0003-3232-0597
Abstract
What is the relative length variance of the luteal phase compared to the follicular phase within healthy, non-smoking, normal-weight, proven normally ovulatory, premenopausal women with normal-length menstrual cycles?
Prospective 1-year data from 53 premenopausal women with two proven normal-length (21–36 days) and normally ovulatory (≥10 days luteal) menstrual cycles upon enrollment showed that, despite 29% of all cycles having incident ovulatory disturbances, within-woman follicular phase length variances were significantly greater than luteal phase length variances.
Many studies report menstrual cycle variability, yet few describe variability in follicular and luteal phase lengths. Luteal lengths are assumed ‘fixed’ at 13–14 days. Most studies have described follicular and luteal phase variability between-women.
This study was a prospective, 1-year, observational cohort study of relative follicular and luteal phase variability both between and within community-dwelling women with two documented normal-length (21–36 days) and normally ovulatory (≥10 days luteal phase) menstrual cycles prior to enrollment. Eighty-one women enrolled in the study and 66 women completed the 1-year study. This study analyzed data from 53 women with complete data for ≥8 cycles (mean 13).
Participants were healthy, non-smoking, of normal BMI, ages 21–41 with two documented normal-length (21–36 days) and normally ovulatory (≥10 days luteal phase) menstrual cycles prior to enrollment. Participants recorded first morning temperature, exercise durations, and menstrual cycle/life experiences daily in the Menstrual Cycle Diary. We analyzed 694 cycles utilizing a twice-validated least-squares Quantitative Basal Temperature method to determine follicular and luteal phase lengths. Statistical analysis compared relative follicular and luteal phase variance in ovulatory cycles both between-women and within-woman. Normal-length cycles with short luteal phases or anovulation were considered to have subclinical ovulatory disturbances (SOD).
The 1-year overall 53-woman, 676 ovulatory cycle variances for menstrual cycle, follicular, and luteal phase lengths were 10.3, 11.2, and 4.3 days, respectively. Median variances within-woman for cycle, follicular, and luteal lengths were 3.1, 5.2, and 3.0 days, respectively. Menstrual cycles were largely of normal lengths (98%) with an important prevalence of SOD: 55% of women experienced >1 short luteal phase (<10 days) and 17% experienced at least one anovulatory cycle. Within-woman follicular phase length variances were greater than luteal phase length variances (P < 0.001). However, follicular (P = 0.008) and luteal phase length (P = 0.001) variances, without differences in cycle lengths, were greater in women experiencing any anovulatory cycles (n = 8) than in women with entirely normally ovulatory cycles (n = 6).
Limitations of this study include the relatively small cohort, that most women were White, initially had a normal BMI, and the original cohort required two normal-length and normally ovulatory menstrual cycles before enrollment. Thus, this cohort’s data underestimated population menstrual cycle phase variances and the prevalence of SOD.
Our results reinforce previous findings that the follicular phase is more variable than the luteal phase in premenopausal women with normal-length and ovulatory menstrual cycles. However, our study adds to the growing body of evidence that the luteal phase is not predictably 13–14 days long.
This medical education project of the University of British Columbia was funded by donations to the Centre for Menstrual Cycle and Ovulation Research. The authors do not have any conflicts of interest to disclose.
N/A.
Introduction
The menstrual cycle is conventionally divided into two main phases: the preovulatory follicular phase (FP) and the post-ovulatory luteal phase (LP). The length of these two phases is determined by a complex interplay of hormone signaling within the hypothalamic–pituitary–gonadal axis and reflects the actions of estradiol and progesterone. The FP starts from the first day of menstrual flow up to the day before ovulation and involves secretion of 17-β-estradiol from granulosa cells, formation of a dominant follicle, and proliferation of the endometrium, followed by an estradiol-peak. This stimulates a spike in LH from the pituitary. The LH peak is usually followed by a rupture of the dominant follicle (ovulation). The LP starts with the day of ovulation and continues until the day before the next menstrual flow and is a critical component of the menstrual cycle for fertility. After ovulation, the corpus luteum begins making high amounts of progesterone as well as estradiol; progesterone transforms the FP’s proliferative endometrium into secretory endometrium in preparation for potential implantation by a blastocyst. If fertilization does not occur, the corpus luteum regresses, levels of estradiol and progesterone fall, and menses begin (Costanzo, 2007).
Physiology and other texts often assert that normal-length menstrual cycles (21–35 days) are consistently ovulatory cycles, and further, that the LP is of stable length at 13–14 days (Presser, 1974; Costanzo, 2007; Hall, 2022). Variability in the length of the menstrual cycle is attributed largely to variability in the length of the FP (Presser, 1974; Costanzo, 2007; Hall, 2022). Indeed, multiple studies show that the mean or median LP length in premenopausal women is 13–14 days (Vollman, 1977; France et al., 1983; Ecochard and Gougeon, 2000; Harlow, 2000). These statements, however, may under-acknowledge the variability in LP lengths both between and within premenopausal women with normal-length menstrual cycles (Lenton et al., 1984a,b; Prior et al., 1990a; Harlow, 2000; Fehring et al., 2006; Cole et al., 2009).
Vollman’s (1977) menstrual cycle monograph showed that the FP and LP lengths were both variable and also that both systematically changed across women’s reproductive lifetimes. Vollman analyzed cycles by the ‘mean basal body temperature method’, which our lab subsequently validated (Prior et al., 1990b) along with the new ‘least squares method’ of Quantitative Basal Temperature© (QBT©) analysis. Vollman recorded a FP length range of 8–29 days across the reproductive years. The FP length is longer and more variable in adolescence, then shortens and becomes more stable in the premenopausal years. It shortens again in early perimenopause and then lengthens and becomes more variable in late perimenopause (Vollman, 1977; Lenton et al., 1984a).
Other studies of premenopausal women with normal-length menstrual cycles that have determined FP length by time to the LH surge have found variable mean FP lengths across different populations: 12.9 days (95% Confidence Interval [95% CI] 8.2–20.5 days) (Lenton et al., 1984a), 14.7 days (SD ± 2.4 days) (Cole et al., 2009), and 16.5 days±3.4 days (Fehring et al., 2006). One study evaluated ovulation by ultrasound in 199 cycles from 80 women with a history of menstrual cycles between 24 and 34 days in length (Ecochard and Gougeon, 2000). This investigation found a mean FP length of 14.59 ± 0.33 days (Ecochard and Gougeon, 2000). Two review papers have attempted to define the normal range of the FP length to be 10–20 days (Abraham, 1978) or 10–23 days (Harlow, 2000), respectively.
LP length also has between-women and within-woman variation and systematically changes across women’s reproductive lifetimes. Vollman determined that the LP can range from zero (in an anovulatory cycle) to 19 days, but that more than 90% of cycles had an LP between 7 and 15 days long by the mean temperature QBT© method (Vollman, 1977). LP lengths are shortest both in the first few years after menarche (Vollman, 1977; Sun et al., 2019) and in perimenopause (Lenton et al., 1984b; Prior, 2019). They are most stable between premenopausal women; however, short LPs (<10 days by QBT©) can and do occur during the premenopausal years (Vollman, 1977; Prior, 2019, 2022).
Studies of premenopausal women with normal-length menstrual cycles that have determined LP lengths by the serum or urine LH surge, have found variable mean lengths across different populations: 14.3 days (95% CI 11.3–17.0 days) (Lenton et al., 1984b), 12.4 ± 2.0 days (Fehring et al., 2006), and 13.2 ± 2.0 days (Cole et al., 2009). By ultrasound, in a cohort of 80 women with strict criteria for the history of normal-length menstrual cycles and no LP defects, the LP length was found to have a mean of 13.64 ± 0.25 days (Ecochard and Gougeon, 2000). A review paper on normal menstrual cycle characteristics has reported LP lengths between women range from 8 to 17 days (Harlow, 2000).
There have been few studies in normal-length menstrual cycles on the variability of the FP and LP lengths within premenopausal women (Fehring et al., 2006; Cole et al., 2009). Determining the variability of LP lengths within regularly cycling premenopausal women is of physiological importance since ovulatory disturbances (short LPs and anovulation) within normal-length cycles have been documented by meta-analysis to associate with spinal bone loss (Li et al., 2014) and in the past were associated with infertility (Jones, 1949). Only one recent study has associated short LP lengths with short-term infertility (Crawford et al., 2017). Luteal length variability may be more common than is currently recognized and thus ‘Subclinical Ovulatory Disturbances’ (SOD) (normal-length, regular cycles with short LPs or anovulation, SOD) may occur commonly (Prior, 2019). The primary objective of our study was to assess the relative degree of variability in FP and LP lengths in ovulatory cycles within individual premenopausal women.
Materials and methods
Study design
This study was a secondary analysis of data collected during a 1-year prospective cohort study of 66 premenopausal women who had two consecutive normally ovulatory and normal-length menstrual cycles (21–36 days) prior to enrollment. The original study (Prior et al., 1990a) was run by Dr Prior, as is this current research, under the auspices of the Centre for Menstrual Cycle and Ovulation Research (CeMCOR) in connection with the Department of Medicine/Endocrinology at the University of British Columbia. The study participants were recruited via advertisement in the community and provided informed, signed, written consent that the study data could be used for future, secondary scientific analysis. The overarching study protocol can be found at: https://open.library.ubc.ca/soa/cIRcle/collections/facultyresearchandpublications/52383/items/1.0406312?o=9.
We analyzed menstrual cycle data that were collected over 1 year from 53 healthy, non-smoking, normal-weight premenopausal women (ages 21–41 years) who had been proven normally menstruating and ovulating on two consecutive cycles prior to study entry and who each contributed ≥8 cycles (termed the Prospective Ovulatory Cohort [POC]).
Our primary objective was the relative degree of variability in FP and LP lengths in ovulatory cycles. Our secondary objectives were to assess and describe: (i) the variances in FP and LP lengths between-women and within-woman and, (ii) the variances in FP and LP lengths within-woman stratified by 1-year ovulation experiences (all normally ovulatory versus presence of SOD with any anovulatory cycles). We hypothesized that in 1-year, prospective data within proven normally ovulatory, healthy, non-smoking, normal-weight premenopausal women, variability of LP lengths may approach the variability that is known to occur in the FP in normal-length (21–36 days) and ovulatory cycles. We hypothesized also that the between-women variance in FP and LP lengths will be greater than the within-woman variance. Lastly, we expected that the FP and LP length variances will be greater in women with anovulatory cycles±short LP cycles than in women with all normally ovulatory cycles over 1 year.
Ethics
The data for analysis included demographic, anthropomorphic, reproduction, and lifestyle data, Menstrual Cycle Diary© data (Prior, 1996) and QBT© data. The original study was approved by the University of British Columbia (UBC) Clinical Research Ethics Board (CREB) (No. C84-007). This secondary analysis of the data is approved by the UBC CREB (No. H20-01937).
Study population
The process of recruitment and screening has been described previously (Prior et al., 1990a) and is illustrated in Fig. 1. Two hundred forty-five women responded to an advertisement in the community to participate in the study. One hundred thirty-two women were excluded on the bases of the exclusion criteria including (i) use of an oral contraceptive, glucocorticoid, or other bone-active drug in the past 6 months, (ii) BMI<18.5 or >25.0 kg/m2, (iii) change in weight of >2.5 kg in the last year, (iv) shift work (which would affect basal temperature), (v) history of an eating disorder, (vi) compulsive exercise or (vii) current smokers. Low-reading mercury thermometers were provided to the remaining 113 participants (Becton Dickinson, No. 4009) and read to the nearest 0.05 °C. Women were asked to record first morning temperatures and menstrual cycle experiences on the Menstrual Cycle Diary© for two cycles. On the basis of two consecutive normal-length (cycle days 21–36) and normally ovulatory (LP lengths ≥10 days) cycles, 81 women were initially enrolled in the study. All women were White, except for two, one Chinese and one Filipino. All were then invited to record first morning temperature at the bottom of the Diary, menstrual cycle and life experiences (optional) and exercise durations daily in the evening over the 1-year study period. They were seen by researchers every 3 months and contacted monthly by telephone.
Flow of participants through the prospective ovulation cohort to study cohort of 53 women who had two normally ovulatory, normal-length menstrual cycles (21–36 days with luteal phase ≥10 days) on enrollment and each contributed ≥8 menstrual cycles of which 694 of 720 cycles had ovulation data.
Sixty-six women completed the 1-year study. Of these, 53 women had completed Menstrual Cycle Diary© (optional) as well as the required QBT© data for ≥8 menstrual cycles (mean 13 cycles/woman). These 53 women provided data for this study.
Menstrual Cycle Diary© and Quantitative Basal Temperature©
Menstrual Cycle Diary© and basal (first morning) temperature data were recorded on paper by the participants for the original study and these data were translated to a digital format by CeMCOR researchers. Cycles were excluded if sleep-wake times were highly irregular (leading to erratic temperature changes), if there was a woman-noted fever mid-cycle, or if ≥33% of the temperatures or any three temperatures at the mid-cycle were missing; few data were excluded. Basal temperature data were analyzed by least mean square QBT© analysis, which marked the largest and most significant (P<0.05) change in basal temperature as the QBT shift day which correlated with the daily mid-cycle serial samples for serum luteinizing hormone (LH) peak (r = 0.879). There was a mean delay in the temperature shift following the LH surge of 2.4 ± 1.5 days (Prior et al., 1990b). The first day of LP onset by the least mean square QBT© method also correlated well with the urine pregnanediol-3-glucuoronide 3-fold follicular-to-luteal increase method (r = 0.803) despite extremely variable wake-sleep times in the participants (Bedford et al., 2009).
The length of the menstrual cycle was determined from the first day of menstrual flow to the day before the onset of the next menstrual flow. The length of the FP was the first day of menstrual flow to the day before the QBT© shift day. The length of the LP was the QBT© shift day to the day before the next menstrual flow. Normal LP lengths were defined as ≥10 days and short LP lengths as <10 days (Vollman, 1977; Prior, 2019). Cycles were deemed anovulatory if there were no statistically significant rise in basal temperature for at least 3 days.
Statistical analysis
All data were analyzed in either IBM® Statistical Package for the Social Sciences (SPSS®) software version 27.0.10. or Microsoft Excel. Data were analyzed in three groups: (i) all women (n = 53), (ii) women with a mean LP length ≥10 days (n = 38), or (iii) women with a mean LP length <10 days (n = 15). Descriptive statistics including mean, SD, 95% CI, median, range, and variance were generated for demographic, anthropomorphic, reproductive, and lifestyle data. Primary analysis of the menstrual cycle data included only ovulatory cycles from the 53 women (total cycles = 676) in order to compare FP and LP lengths. We also described the analysis of SOD. Anovulatory cycles were arbitrarily given a luteal length of 0.1 days. Descriptive statistics were generated as above for each of menstrual cycle, FP and LP lengths, and their respective variances both between-women and within-woman. The median FP:LP length ratio was calculated for ovulatory cycles within each woman’s total 1-year data.
All data were assessed for normal distribution using the Shapiro–Wilk test and were reported as either means and 95% CI (for normally distributed data) or median and range (for non-normally distributed data). Statistically appropriate tests (independent sample t-test for data with a normal distribution and Mann–Whitney U test for data with a non-normal distribution) were run to compare the means or medians of these parameters. The relative variance of the FP and LP lengths within woman were then analyzed using the non-parametric Wilcoxon signed rank test in SPSS®. Women with entirely normally ovulatory menstrual cycles over 1-year (n = 6) versus women with normally ovulatory, short luteal, and any anovulatory cycles (n = 9) were compared between women for the variances of FP to LP lengths. Standard box and whisker plots were used to illustrate some of the data. They showed a box bordered by the 25th and 75th percentiles with a median line plus whiskers indicating the minimum and maximum data. We used multiple linear regression for FP evaluation in relationship to baseline variables; stepwise negative regression was used to assess LP length categories. A double-sided P-value of 0.05 or less was considered important.
Results
Menstrual cycle data
Figure 1 shows the flow of participants from recruitment to the formation of this study cohort which included 53 healthy, community-dwelling women with two, consecutive, normal-length (21–36 days; Abraham, 1978), and ovulatory menstrual cycles upon enrollment. The 53 women in this analysis had a mean age of 33.9 years (baseline range 21–41 years), a normal BMI of 22.0 kg/m2, and age of menarche of 11.5 years (Table 1). Median past combined hormonal contraceptive use was 33.0 months (range 0–156 months). Median past months of pregnancy was 3.0 (range 0–40 months) and percent ever pregnant was 54.7%. This cohort was quite physically active with many being recreational runners with the mean minutes of moderate to intense physical activity per cycle being 82.2. There was no statistically significant difference in anthropometric, demographic, reproductive, and lifestyle data when women were stratified by their mean 1-year LP length.
Demographic, anthropometric, reproductive, and menstrual cycle characteristics of the prospective ovulation cohort by mean luteal phase lengths (LP)a in the normal (≥10 days) or short (<10 days) range.
| All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-value | |
|---|---|---|---|---|
| Total number of ovulatory cycles | 676 | 485 | 191 | |
| bMean (95% confidence interval) | ||||
| Age (years) | 33.9 (32.4; 35.5) | 33.6 (31.9; 35.4) | 34.7 (31.4; 38.0) | 0.545 |
| Height (cm) | 162.5 (160.8; 164.3) | 162.4 (160.4; 164.4) | 162.8 (158.7; 166.9) | 0.837 |
| Weight (kg) | 58.1 (56.2; 59.9) | 58.6 (56.5; 60.7) | 56.9 (52.7; 61.1) | 0.417 |
| BMI (kg/m2) | 22.0 (21.4; 22.5) | 22.2 (21.6; 22.8) | 21.5 (20.0; 22.9) | 0.258 |
| Age at menarche (years) | 11.5 (11.1; 11.8) | 11.4 (11.0; 11.7) | 11.7 (11.0; 12.4) | 0.408 |
| cMedian (min–max) | ||||
| Average minutes of moderate-vigorous exercise (min/cycle) | 82 (0–854) | 83 (0–477) | 59 (0–854) | 0.678 |
| Past months on combined hormonal contraception | 33 (0–156) | 27 (0–143) | 39 (0–156) | 0.096 |
| % ever pregnant | 54.7% | 52.6% | 60.0% | 0.627 |
| All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-value | |
|---|---|---|---|---|
| Total number of ovulatory cycles | 676 | 485 | 191 | |
| bMean (95% confidence interval) | ||||
| Age (years) | 33.9 (32.4; 35.5) | 33.6 (31.9; 35.4) | 34.7 (31.4; 38.0) | 0.545 |
| Height (cm) | 162.5 (160.8; 164.3) | 162.4 (160.4; 164.4) | 162.8 (158.7; 166.9) | 0.837 |
| Weight (kg) | 58.1 (56.2; 59.9) | 58.6 (56.5; 60.7) | 56.9 (52.7; 61.1) | 0.417 |
| BMI (kg/m2) | 22.0 (21.4; 22.5) | 22.2 (21.6; 22.8) | 21.5 (20.0; 22.9) | 0.258 |
| Age at menarche (years) | 11.5 (11.1; 11.8) | 11.4 (11.0; 11.7) | 11.7 (11.0; 12.4) | 0.408 |
| cMedian (min–max) | ||||
| Average minutes of moderate-vigorous exercise (min/cycle) | 82 (0–854) | 83 (0–477) | 59 (0–854) | 0.678 |
| Past months on combined hormonal contraception | 33 (0–156) | 27 (0–143) | 39 (0–156) | 0.096 |
| % ever pregnant | 54.7% | 52.6% | 60.0% | 0.627 |
Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
ANOVA.
Data that did not follow a normal distribution with Shapiro-Wilk test were compared with Mann-Whitney U test.
Demographic, anthropometric, reproductive, and menstrual cycle characteristics of the prospective ovulation cohort by mean luteal phase lengths (LP)a in the normal (≥10 days) or short (<10 days) range.
| All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-value | |
|---|---|---|---|---|
| Total number of ovulatory cycles | 676 | 485 | 191 | |
| bMean (95% confidence interval) | ||||
| Age (years) | 33.9 (32.4; 35.5) | 33.6 (31.9; 35.4) | 34.7 (31.4; 38.0) | 0.545 |
| Height (cm) | 162.5 (160.8; 164.3) | 162.4 (160.4; 164.4) | 162.8 (158.7; 166.9) | 0.837 |
| Weight (kg) | 58.1 (56.2; 59.9) | 58.6 (56.5; 60.7) | 56.9 (52.7; 61.1) | 0.417 |
| BMI (kg/m2) | 22.0 (21.4; 22.5) | 22.2 (21.6; 22.8) | 21.5 (20.0; 22.9) | 0.258 |
| Age at menarche (years) | 11.5 (11.1; 11.8) | 11.4 (11.0; 11.7) | 11.7 (11.0; 12.4) | 0.408 |
| cMedian (min–max) | ||||
| Average minutes of moderate-vigorous exercise (min/cycle) | 82 (0–854) | 83 (0–477) | 59 (0–854) | 0.678 |
| Past months on combined hormonal contraception | 33 (0–156) | 27 (0–143) | 39 (0–156) | 0.096 |
| % ever pregnant | 54.7% | 52.6% | 60.0% | 0.627 |
| All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-value | |
|---|---|---|---|---|
| Total number of ovulatory cycles | 676 | 485 | 191 | |
| bMean (95% confidence interval) | ||||
| Age (years) | 33.9 (32.4; 35.5) | 33.6 (31.9; 35.4) | 34.7 (31.4; 38.0) | 0.545 |
| Height (cm) | 162.5 (160.8; 164.3) | 162.4 (160.4; 164.4) | 162.8 (158.7; 166.9) | 0.837 |
| Weight (kg) | 58.1 (56.2; 59.9) | 58.6 (56.5; 60.7) | 56.9 (52.7; 61.1) | 0.417 |
| BMI (kg/m2) | 22.0 (21.4; 22.5) | 22.2 (21.6; 22.8) | 21.5 (20.0; 22.9) | 0.258 |
| Age at menarche (years) | 11.5 (11.1; 11.8) | 11.4 (11.0; 11.7) | 11.7 (11.0; 12.4) | 0.408 |
| cMedian (min–max) | ||||
| Average minutes of moderate-vigorous exercise (min/cycle) | 82 (0–854) | 83 (0–477) | 59 (0–854) | 0.678 |
| Past months on combined hormonal contraception | 33 (0–156) | 27 (0–143) | 39 (0–156) | 0.096 |
| % ever pregnant | 54.7% | 52.6% | 60.0% | 0.627 |
Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
ANOVA.
Data that did not follow a normal distribution with Shapiro-Wilk test were compared with Mann-Whitney U test.
In total, there were 720 cycles recorded by the 53 women. Twenty-six cycles were excluded on the basis of incomplete or unanalyzable QBT© data. Of the 694 remaining cycles, 676 cycles (97.4%) were ovulatory and 18 cycles (2.6%) were anovulatory. There was no statistically significant difference in menstrual cycle length between ovulatory and anovulatory cycles (median 28 vs 29 days, P = 0.999) when all women’s data were examined (Fig. 2).
Comparison of menstrual cycle lengths in all ovulatory (n = 676) and all anovulatory (n = 18) cycles over 1 year by box and whisker plots in 53 premenopausal women in the Prospective Ovulation Cohort. The line within the box is the median with the lower border being the 25th and the upper the 75th percentile of the data. The extremes of the whiskers show the data’s minimum and maximum values. Outliers are designated by superscript o and an extreme outlier by superscript *. Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
In all entirely ovulatory cycles (n = 676), the mean number of ovulatory cycles was 12.8 cycles/woman. Median menstrual cycle length was 28.1 days (95% CI range 27.5–28.8 days) with 2.4% of cycles outside the normal 21–36 day length. Mean FP length was 17.6 days (95% CI 17.0–18.2) and median LP length was 10.9 days (range 8.3–12.5 days). Median FP:LP length ratio within-woman was 1.7 (range 1.2–2.6) (Table 2).
Between-women analysis of cycle number, averages of menstrual cycle, follicular phase, and luteal phase (LP) lengths in all ovulatory cycles in the Prospective Ovulation Cohort and also as stratified by mean LP length (≥10 days; ˂10 days).a
| All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-value | |
|---|---|---|---|---|
| Total number of ovulatory cycles | 676 | 485 | 191 | |
| bMean (95% confidence interval) | ||||
| Number of cycles | 12.8 (11.8; 13.7) | 12.8 (11.8; 13.7) | 12.7 (10.4; 15.1) | 0.977 |
| Cycle length (days) | 28.1 (27.5; 28.8) | 28.3 (27.7; 29.0) | 27.7 (26.1; 29.3) | 0.362 |
| Follicular length (days) | 17.6 (17.0; 18.2) | 17.2 (16.6; 17.8) | 18.6 (17.2; 20.1) | 0.032 |
| cMedian (min–max) | ||||
| Luteal length (days) | 10.9 (8.3–12.5) | |||
| Follicular:luteal phase length ratio | 1.7 (1.2–2.6) | |||
| All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-value | |
|---|---|---|---|---|
| Total number of ovulatory cycles | 676 | 485 | 191 | |
| bMean (95% confidence interval) | ||||
| Number of cycles | 12.8 (11.8; 13.7) | 12.8 (11.8; 13.7) | 12.7 (10.4; 15.1) | 0.977 |
| Cycle length (days) | 28.1 (27.5; 28.8) | 28.3 (27.7; 29.0) | 27.7 (26.1; 29.3) | 0.362 |
| Follicular length (days) | 17.6 (17.0; 18.2) | 17.2 (16.6; 17.8) | 18.6 (17.2; 20.1) | 0.032 |
| cMedian (min–max) | ||||
| Luteal length (days) | 10.9 (8.3–12.5) | |||
| Follicular:luteal phase length ratio | 1.7 (1.2–2.6) | |||
Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
ANOVA.
Data that did not follow a normal distribution with Shapiro-Wilk test were compared with Mann-Whitney U test.
Bold P value indicates a significant one.
Between-women analysis of cycle number, averages of menstrual cycle, follicular phase, and luteal phase (LP) lengths in all ovulatory cycles in the Prospective Ovulation Cohort and also as stratified by mean LP length (≥10 days; ˂10 days).a
| All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-value | |
|---|---|---|---|---|
| Total number of ovulatory cycles | 676 | 485 | 191 | |
| bMean (95% confidence interval) | ||||
| Number of cycles | 12.8 (11.8; 13.7) | 12.8 (11.8; 13.7) | 12.7 (10.4; 15.1) | 0.977 |
| Cycle length (days) | 28.1 (27.5; 28.8) | 28.3 (27.7; 29.0) | 27.7 (26.1; 29.3) | 0.362 |
| Follicular length (days) | 17.6 (17.0; 18.2) | 17.2 (16.6; 17.8) | 18.6 (17.2; 20.1) | 0.032 |
| cMedian (min–max) | ||||
| Luteal length (days) | 10.9 (8.3–12.5) | |||
| Follicular:luteal phase length ratio | 1.7 (1.2–2.6) | |||
| All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-value | |
|---|---|---|---|---|
| Total number of ovulatory cycles | 676 | 485 | 191 | |
| bMean (95% confidence interval) | ||||
| Number of cycles | 12.8 (11.8; 13.7) | 12.8 (11.8; 13.7) | 12.7 (10.4; 15.1) | 0.977 |
| Cycle length (days) | 28.1 (27.5; 28.8) | 28.3 (27.7; 29.0) | 27.7 (26.1; 29.3) | 0.362 |
| Follicular length (days) | 17.6 (17.0; 18.2) | 17.2 (16.6; 17.8) | 18.6 (17.2; 20.1) | 0.032 |
| cMedian (min–max) | ||||
| Luteal length (days) | 10.9 (8.3–12.5) | |||
| Follicular:luteal phase length ratio | 1.7 (1.2–2.6) | |||
Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
ANOVA.
Data that did not follow a normal distribution with Shapiro-Wilk test were compared with Mann-Whitney U test.
Bold P value indicates a significant one.
When women were stratified into those with a mean 1-year LP length ≥10 days (n = 38) and <10 days (n = 15) (Table 2) there was no statistically significant difference in a number of cycles/woman. However, there was now a statistically significant difference in mean FP lengths (17.2 versus 18.6 days, P<0.05). Median LP lengths (11.0 and 8.9 days) were obviously significantly different, thus not shown in Table 2, P<0.01.
Between-women and within-woman variance
Across all 53 women in 676 ovulatory cycles, between-women variances for menstrual cycle, FP, and LP lengths were 10.3, 11.2, and 4.3 days, respectively (Table 3). Within-woman variances (also in 676 ovulatory cycles in 53 women) for these three parameters were importantly less than the between-woman variances with median values of 3.1 days (range 0.25–27.25 days), 5.2 days (range 0.5–43.5 days), and 3.0 days (range 0.7–7.9 days), respectively. However, in women with a mean LP length ≥10 days versus those with <10 days, there was no statistically significant difference in within-woman length variances of menstrual cycles, FP, and LP data.
Variances of menstrual cycle, follicular phase, and luteal phase lengths between-women and in all ovulatory cycles in the Prospective Ovulation Cohort and as stratified by mean luteal phase length ≥10 days versus ˂10 days.a
| All ovulatory cycles of 53 women | All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-valueb | |
|---|---|---|---|---|---|
| Number of ovulatory cycles | 676 | 676 | 485 | 191 | |
| Variances median (min–max) | |||||
| Cycle length (days) | 10.3 | 3.1 (0.25–27.3) | 2.9 (0.50–27.3) | 6.2 (0.25–26.1) | 0.353 |
| Follicular length (days) | 11.2 | 5.2 (0.50–43.5) | 4.3 (1.3–43.5) | 6.7 (0.50–31.2) | 0.228 |
| Luteal length (days) | 4.3 | 3.0 (0.70–7.9) | 2.9 (0.70–7.5) | 3.5 (2.0–7.9) | 0.108 |
| All ovulatory cycles of 53 women | All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-valueb | |
|---|---|---|---|---|---|
| Number of ovulatory cycles | 676 | 676 | 485 | 191 | |
| Variances median (min–max) | |||||
| Cycle length (days) | 10.3 | 3.1 (0.25–27.3) | 2.9 (0.50–27.3) | 6.2 (0.25–26.1) | 0.353 |
| Follicular length (days) | 11.2 | 5.2 (0.50–43.5) | 4.3 (1.3–43.5) | 6.7 (0.50–31.2) | 0.228 |
| Luteal length (days) | 4.3 | 3.0 (0.70–7.9) | 2.9 (0.70–7.5) | 3.5 (2.0–7.9) | 0.108 |
Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
Mann–Whitney U test (data that did not follow a normal distribution with Shapiro–Wilk test were compared with a non-parametric test).
Variances of menstrual cycle, follicular phase, and luteal phase lengths between-women and in all ovulatory cycles in the Prospective Ovulation Cohort and as stratified by mean luteal phase length ≥10 days versus ˂10 days.a
| All ovulatory cycles of 53 women | All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-valueb | |
|---|---|---|---|---|---|
| Number of ovulatory cycles | 676 | 676 | 485 | 191 | |
| Variances median (min–max) | |||||
| Cycle length (days) | 10.3 | 3.1 (0.25–27.3) | 2.9 (0.50–27.3) | 6.2 (0.25–26.1) | 0.353 |
| Follicular length (days) | 11.2 | 5.2 (0.50–43.5) | 4.3 (1.3–43.5) | 6.7 (0.50–31.2) | 0.228 |
| Luteal length (days) | 4.3 | 3.0 (0.70–7.9) | 2.9 (0.70–7.5) | 3.5 (2.0–7.9) | 0.108 |
| All ovulatory cycles of 53 women | All women, n = 53 | Mean LP ≥10 days, n = 38 | Mean LP <10 days, n = 15 | P-valueb | |
|---|---|---|---|---|---|
| Number of ovulatory cycles | 676 | 676 | 485 | 191 | |
| Variances median (min–max) | |||||
| Cycle length (days) | 10.3 | 3.1 (0.25–27.3) | 2.9 (0.50–27.3) | 6.2 (0.25–26.1) | 0.353 |
| Follicular length (days) | 11.2 | 5.2 (0.50–43.5) | 4.3 (1.3–43.5) | 6.7 (0.50–31.2) | 0.228 |
| Luteal length (days) | 4.3 | 3.0 (0.70–7.9) | 2.9 (0.70–7.5) | 3.5 (2.0–7.9) | 0.108 |
Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
Mann–Whitney U test (data that did not follow a normal distribution with Shapiro–Wilk test were compared with a non-parametric test).
Results of the Wilcoxon signed rank test comparing FP to LP length variance within-woman showed the FP variance as greater than the LP variance in all ovulatory cycles of 53 women, in normally ovulatory cycles only (P<0.001) and in short LP cycles only (P<0.001) (Table 4). FP length variance was also significantly (P<0.001) greater than LP variance among the 47 women who had both normally ovulatory and short LP cycles. Therefore, we can reject the null hypothesis that the LP variance is the same as the FP variance in healthy, premenopausal women with normal-length and ovulatory menstrual cycles.
Variance of within-women comparison of follicular (FP) and luteal phases (LP) in the Prospective Ovulation Cohort.a
| Follicular phase variance | Luteal phase variance | |||
|---|---|---|---|---|
| Characteristic | Median (min–max) | Mean positive ranks | P-valueb | |
| All ovulatory cycles (676 cycles) in 53 women | 5.2 (0.5–43.5) | 3.0 (0.7–7.9) | 30.5 | <0.001 |
| Normally ovulatory (NOv) (LP≥10 days) cycles (495 cycles) in 53 women | 3.5 (0.0–33.8) | 1.4 (0.0–4.8) | 28.6 | <0.001 |
| Both NOv and SLP cycles (601 cycles) in 47 women (who had both cycle types) | 5.5 (0.5–43.5) | 3.4 (0.8–7.9) | 27.8 | <0.001 |
| Short luteal (LP<10 days) phase (SLP) cycles (181 cycles) in 47 women | 2.0 (0.0–36.8) | 0.5 (0.0–9.0) | 18.9 | <0.001 |
| Follicular phase variance | Luteal phase variance | |||
|---|---|---|---|---|
| Characteristic | Median (min–max) | Mean positive ranks | P-valueb | |
| All ovulatory cycles (676 cycles) in 53 women | 5.2 (0.5–43.5) | 3.0 (0.7–7.9) | 30.5 | <0.001 |
| Normally ovulatory (NOv) (LP≥10 days) cycles (495 cycles) in 53 women | 3.5 (0.0–33.8) | 1.4 (0.0–4.8) | 28.6 | <0.001 |
| Both NOv and SLP cycles (601 cycles) in 47 women (who had both cycle types) | 5.5 (0.5–43.5) | 3.4 (0.8–7.9) | 27.8 | <0.001 |
| Short luteal (LP<10 days) phase (SLP) cycles (181 cycles) in 47 women | 2.0 (0.0–36.8) | 0.5 (0.0–9.0) | 18.9 | <0.001 |
Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
Wilcoxon signed rank test.
Nov, Normally Ovulatory with LP length ≥10 days by QBT; SLP, short luteal phase.
Variance of within-women comparison of follicular (FP) and luteal phases (LP) in the Prospective Ovulation Cohort.a
| Follicular phase variance | Luteal phase variance | |||
|---|---|---|---|---|
| Characteristic | Median (min–max) | Mean positive ranks | P-valueb | |
| All ovulatory cycles (676 cycles) in 53 women | 5.2 (0.5–43.5) | 3.0 (0.7–7.9) | 30.5 | <0.001 |
| Normally ovulatory (NOv) (LP≥10 days) cycles (495 cycles) in 53 women | 3.5 (0.0–33.8) | 1.4 (0.0–4.8) | 28.6 | <0.001 |
| Both NOv and SLP cycles (601 cycles) in 47 women (who had both cycle types) | 5.5 (0.5–43.5) | 3.4 (0.8–7.9) | 27.8 | <0.001 |
| Short luteal (LP<10 days) phase (SLP) cycles (181 cycles) in 47 women | 2.0 (0.0–36.8) | 0.5 (0.0–9.0) | 18.9 | <0.001 |
| Follicular phase variance | Luteal phase variance | |||
|---|---|---|---|---|
| Characteristic | Median (min–max) | Mean positive ranks | P-valueb | |
| All ovulatory cycles (676 cycles) in 53 women | 5.2 (0.5–43.5) | 3.0 (0.7–7.9) | 30.5 | <0.001 |
| Normally ovulatory (NOv) (LP≥10 days) cycles (495 cycles) in 53 women | 3.5 (0.0–33.8) | 1.4 (0.0–4.8) | 28.6 | <0.001 |
| Both NOv and SLP cycles (601 cycles) in 47 women (who had both cycle types) | 5.5 (0.5–43.5) | 3.4 (0.8–7.9) | 27.8 | <0.001 |
| Short luteal (LP<10 days) phase (SLP) cycles (181 cycles) in 47 women | 2.0 (0.0–36.8) | 0.5 (0.0–9.0) | 18.9 | <0.001 |
Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
Wilcoxon signed rank test.
Nov, Normally Ovulatory with LP length ≥10 days by QBT; SLP, short luteal phase.
Subclinical ovulatory disturbances
In reporting on FP and LP relative lengths, it is illustrative to examine them based on the extremes of women’s whole-year ovulatory experience during this study. Of the 694 total cycles from all 53 women, 181 cycles (26.1%) had an LP length <10 days (short LP cycles) and 18 cycles (2.6%) were anovulatory. In ovulatory cycles (n = 676), LP lengths ranged from 3.0 to 16.0 days (median 11.0 days) (Fig. 3).
Luteal phase length frequency distribution in all ovulatory cycles (n = 676) over about 1 year in 53 premenopausal women in the Prospective Ovulation Cohort. Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
The distribution of women’s 1-year ovulatory experiences was as follows: six women (11.3%) had all normally ovulatory cycles, nine women (17.0%) had all ovulatory except for one short LP cycle, 29 women (54.7%) had all ovulatory cycles and more than one short LP cycle, and nine women (17.0%) had one or more anovulatory cycles also with any number of normally ovulatory and short LP cycles.
By giving an anovulatory cycle a ‘LP length’ of 0.1 days, we could compare the variances of FP and LP lengths between these two groups. The median (range) variance of the FP length of the six women with entirely normally ovulatory cycles was 2.4 (2.0–16.2) compared with 15.3 (7.8–35.3) in the eight women with any anovulatory cycles (note that we removed one participant with a majority of anovulatory cycles whose datum was an extreme outlier). The median (range) variance of the LP length in the six women with normally ovulatory cycles was 1.4 (range 0.7–2.0), but the same variance in the eight women with any anovulatory cycles was 13.9 (range 7.8–22.6) (Fig. 4). The differences between these two groups of women were significant for both the variances of FP (P = 0.008) and LP (P = 0.001) lengths.
Variances between women of the follicular and luteal phase (FP and LP) lengths in the six women within the Prospective Ovulation Cohort who, over about 1 year, experienced all normally ovulatory menstrual cycles versus the same variances in the eight women (one outlier excluded) who experienced any anovulatory as well as normal (≥10 days) and short luteal phase ovulatory cycles. The line within the box is the median with the lower border being the 25th and the upper the 75th percentile of the data. The extremes of the whiskers show the data’s minimum and maximum values. Luteal phase lengths of ≥10 days were normal based on least-squares Quantitative Basal Temperature (LS-QBT©) analysis of a first-morning significant temperature increase that occurred 2.4 days following the serum LH peak (Prior et al., 1990b).
We evaluated the relationship of the FP length to baseline variables using multivariable linear regression which showed that the only important variable (P = 0.005) was age (B = −0.152) which means that for each year older, the FP shortened by .152 days after adjustment for BMI, mod-vigorous exercise, age at menarche, months of pregnancy, and months on CHC. For assessment of the LP length (as ≥10 or <10 days) versus baseline variables we used a backward stepwise multivariable regression. The variance in step 2 was 15.2% and the OR for ‘Months on CHC’ was 0.986 (P = 0.047). This means that for every additional month of CHC use, the odds of a normally ovulatory menstrual cycle decreased by about 1.4%.
Discussion
This 1-year prospective study described the frequency and variability of menstrual cycle, FP, and LP lengths both between-women and within-woman showing a significantly greater between-women variance for all. Most importantly, it documented that, even in this cohort with careful prospective LP length documentation, and the majority of women who had cycles with SOD, the FP length remained significantly more variable than the LP length. These data are in a cohort of 53 healthy, community-dwelling, premenopausal women who were proven to have two normal-length (21–36 days) and normally ovulatory (LL ≥10 days) cycles by the least-squares QBT© method prior to enrollment. In this discussion, we will focus on comparing the literature first on our results between-women and then compare within-woman data with previous studies that have similarly attempted to demonstrate the normal range and variability of the menstrual cycle and its two phases.
Range of FP and LP lengths
The median menstrual cycle length of 28.1 days (95% CI 27.5–28.8 days, 95% range 23–36 days) in our cohort is consistent with other previous studies (Vollman, 1977; Abraham, 1978; Landgren et al., 1980; Ecochard and Gougeon, 2000; Fehring et al., 2006; Cole et al., 2009; Li et al., 2023). Our results on FP and LP lengths are consistent with Vollman’s cross-sectional results in cycles of 21–36 days. His data from premenopausal women show a FP range of 8–29 days and LP range of 4–16 days, with 90% of LP lengths between 7 and 15 days, as compared in our cohort to the median FP length of 17 days (range 8–36 days) and LP length of 11 days (range 3–16 days).
In the 1980s, Lenton et al. (1984a,b) analyzed 295 normal-length (<40 days) and ovulatory menstrual cycles from 205 premenopausal women (ages 19–39) using the serum LH peak. The mean FP length was 12.9 days (95% range 8.2–20.5 days), which is shorter than our mean FP length of 17.6 days (95% range 12–26 days). Comparison of LP length between our study and the Lenton studies is not possible, however, since the Lenton studies excluded short LP cycles when reporting the mean and 95% CI of LP length. However, the Lenton studies did find an LP range of 7–18 days with premenopausal women having 2.9–14.4% of cycles with short LPs (<12 days by serum LH peak). Thus, Lenton also showed that the LP in ovulatory cycles is not a constant 13–14 days.
Fehring et al. (2006) studied 1060 menstrual cycles from 141 healthy, premenopausal women (ages 21–44) with a history of 21- to 42-day-long menstrual cycles. Each woman tracked urinary hormones with a fertility monitor and contributed between 3 and 13 menstrual cycles. FP and LP lengths were determined using the urine LH surge. They had similar results with only a slightly shorter mean and 95% CI range for the FP length of 16.5 days (95% CI range 9–23 days) and a slightly longer mean LP length of 12.4 days (95% CI range 8–17 days) (Fehring et al., 2006).
Cole et al. (2009) studied 405 non-conceptive cycles from 184 premenopausal women with no history of infertility or menstrual cycle disorders. They measured urine LH surge and, in comparison to our results, found a slightly shorter mean FP length of 14.7 ± 2.4 days (95% range 10–20 days) and slightly longer mean LP length of 13.2 ± 2.0 days (95% range 9–17 days).
More recently, Bull et al. (2019) studied 612 613 ovulatory cycles from 24 648 anonymized women who used a menstrual cycle tracking app and contributed BBT measurements. Unlike our study, the Bull et al. study included cycles up to 90 days (<1% of cycles were greater than 50 days). They thus had a slightly longer mean menstrual cycle length of 29.3 ± 5.2 days. Mean follicular length was slightly shorter at 16.9 ± 5.3 days and mean LP length slightly longer at 12.4 ± 2.4 days. The Bull et al. paper reported that both mean FP and LP lengths increased with increasing cycle length.
These small differences are a reasonable variation between populations and the different methods of determining FP and LP lengths. As the LH surge occurs a few days before the post-ovulatory progesterone-driven basal temperature rise, the FP lengths in our study will naturally be a bit longer and the LP lengths a bit shorter than those reported by studies using the serum LH peak or urine LH surge to determine LP length. Our LS-QBT method to define the QBT shift day has been shown to correlate with the serum LH peak (r = 0.879) with a mean delay of 2.4 ± 1.5 days (Prior et al., 1990b). Vollman also used a QBT method to determine FP and LP lengths and found similar FP and LP ranges to our study, albeit with a slightly longer mean LP (11.4–12.4 days) which may be due to the inclusion of cycles up to 40 days long (Vollman, 1977).
Overall, in normal-length and ovulatory menstrual cycles, FP, and LP lengths appear to generally range between 12–26 days and 4–16 days, respectively, by QBT© methods and between 9–23 days and 8–17 days by urine LH surge methods.
Between and within-woman variability of FP and LP lengths
Few studies have reported on the variability of the FP and LP lengths between-women and fewer still on the variability of these phases within-woman. In this study, the variability of the menstrual cycle, FP and LP lengths between women were 10.3, 11.2, and 4.3 days, respectively, which corroborates previous findings that the FP is more variable than the LP between women (Presser, 1974; Hall, 2022). Our results are more variable than those of Cole et al. (2009) who reported menstrual cycle, FP, and LP variability between women of 5.2, 6.6, and 4.0 days, respectively. This may be because these authors excluded abnormal cycles with ‘extraordinary hormone results and cycle timings’ plus they recorded cycles over a shorter time interval.
Our study showed that between-women variance was greater than within-woman variance for all menstrual cycle parameters, which is similar to the findings by Cole et al. (2009). Median menstrual cycle length variance was 3.1 days (range 0.25–27.25 days) within-woman, which is in line with the updated FIGO recommendation of ‘normal’ cycle variability of 7 days (i.e. ± 4 days) (Munro et al., 2018). Our median FP and LP length variances of 5.2 days (range 0.5–43.5) and 3.0 days (range 0.7–7.9), respectively, were greater than that reported by Cole et al. (2009) (mean ± SD of 3.9 ± 3.7 days and 2.6 ± 3.1 days, respectively). Fehring et al. (2006) also found less variability within-woman than in our study with FP and LP length variability of greater than 7 days to be 33.6% and 9.0%, respectively (versus 50.9% and 33.6% in this study). The reasons for this greater variance in both FP and LP lengths in our study compared to those by Cole et al. and Fehring et al. are unclear, but again, may be due to differences in a number of menstrual cycles recorded per woman, different methods to determine FP and LP lengths and perhaps the exercise patterns of participants in our study.
Our results, that FP variance was greater than LP variance in all ovulatory cycles, are in alignment with results from Cole et al. (2009) and Fehring et al. (2006) and with our current understanding of menstrual cycle variability. However, our result of a median luteal length variance within-woman of 3.0 days (range of 0.7–7.9 days) highlights that, although less than the variability of the FP, there is also significant variability in the length of the LP. Fehring et al. (2006) similarly reported that the LP contributes ∼25% to the intra-cycle length differences. When women were stratified by mean LP length, women with mean LP <10 days had more within-woman LP variance [median 3.5 days (range 2.0–7.9)] than those with mean LP ≥10 days [median 2.9 days (range 0.70–7.5)] (P = 0.108). Although these results are not statistically significant, it shows that those women who experience short LP cycles still exhibit variance in the LP and usually have a mix of normal and short LP cycles.
These current results contribute to a growing body of evidence that the LP has variable lengths. They also challenge the assertion that the LP is a predictable 13–14 days in normal-length, ovulatory menstrual cycles from healthy, premenopausal women. Our findings have implications for individuals who are either trying to avoid or achieve pregnancy using the calendar method. In those women with normal-length menstrual cycles, a LP length of 13–14 days cannot be assumed in order to determine a woman’s average FP length and thus the estimated day of ovulation and the ‘fertility window’. These data also reveal the inaccuracy of cycle day or backward cycle day counting as a method with which to determine “ovulation” and the “LP” is used today in many menstrual cycle apps and publications.
Subclinical ovulatory disturbances
Normal-length cycles with short LP lengths or anovulation, called SOD (Prior, 2019), were common in the 53 proven normally cycling and ovulatory women with 54.7% having multiple short LP lengths and 17.0% having at least one anovulatory cycle during the approximately 1-year study period. Only six women had consistently ovulatory and normal LP length cycles.
The fact that there was no significant difference between the median menstrual cycle lengths in ovulatory versus anovulatory cycles (28. versus 29 days, P = th0.999) calls into question the concept that normal-length menstrual cycles are always ovulatory (Hall, 2022). Prospective data from premenopausal women with normal-length menstrual cycles have shown that the number of women who experience ≥1 short LP cycle can range from 13% to 82% and the number of women who experience ≥1 anovulatory cycle can range from 3% to 61% in a study population (Prior et al., 1990a, 2015; Waugh et al., 2007; Bedford et al., 2010; Schliep et al., 2014). The large differences in percentages between study populations are likely due to differences in lengths of time each woman was studied, the age and gynecological age of the participants, the specificity of methods for determining the LP length, and other, often unmeasured, variables.
From a physiological perspective, SOD are likely related to lower mean LP progesterone levels or integrated progesterone exposures (Schliep et al., 2014; Prior et al., 2015; Francis and Keay, 2023) as the Francis and Keay data now confirm. This information is clinically important because in meta-analysis the prevalence of SOD has been significantly associated with spinal bone mineral density loss (Li et al., 2014) and may be associated with more difficulty attaining or maintaining a pregnancy in otherwise healthy women with normal-length menstrual cycles (Horta et al., 1977; Crawford et al., 2017). A recent opinion paper postulated that SOD may be part of a spectrum of normal, adaptive, reversible, reproductive changes in response to physical, psychological, social, or environmental stressors (Prior, 2022). The author described these reproductive changes using an iceberg analogy, wherein the clinically apparent, but less common, amenorrhea and oligomenorrhea lie above the water and are observable, while the less clinically obvious, but seemingly more common, SOD lie under-recognized below the water (Prior, 2022). This perspective may lead to a better understanding of subtle physiologic and ovulatory reproductive adaptations in premenopausal women who still have regular and normal-length menstrual cycles leading to further research on their significance for women’s reproductive and overall health.
Strengths of this study include the study length of approximately 1 year, with a mean of 13 cycles recorded per woman, which was likely long enough to capture a representative picture of the typical individual variances in cycle, FP, and LP lengths. It also allowed the documentation of incident SOD. This cohort comprised healthy, non-smoking women with a normal BMI although some women trained for and ran a marathon during the study but had no difference in cycle characteristics. Furthermore, the LS-QBT© method utilized in this study to determine FP and LP lengths is a multiply-validated, quantitative method that correlates well with the serum LH peak (r = 0.879) which is a commonly used standard for indirect assessment of ovulation (Lenton et. al. 1984b) (Prior et al., 1990b).
Our study is limited by the fact that this is a secondary analysis of a relatively small cohort of 53 women. From the original 66 women, we selected 53 women who contributed comprehensive data on eight or more menstrual cycles. There were, however, no statistically significant differences in demographic, anthropometric, or menstrual cycle parameters between women who were included vs those who were excluded from this analysis (n = 9). We also only knew months of pregnancy but not the parity of participants; it did not play a significant factor in this analysis. The inclusion criteria for the original study (Prior et al., 1990a) required women to have two normal-length and ovulatory menstrual cycles before they could be enrolled. That would have excluded women with more frequent ovulatory disturbances; thus, results from this cohort likely underestimate the actual prevalence of SOD, as well as the degree of individual variance in menstrual cycle phases. Our study excluded participants with a BMI ≥25.0 kg/m2. Recent studies using menstrual cycle apps have shown a correlation of higher BMI with longer menstrual cycles, longer FP lengths, and shorter LP lengths (Bull et al., 2019; Li et al., 2023). More research is needed to further describe both between women, and within-woman menstrual cycle phase variabilities by BMI. Finally, it is a limitation that our cohort comprised mostly White women. A recent study on the menstrual cycle variation by demographic characteristics of women residing in the USA and using a menstrual cycle app found that, on average Asian women had longer menstrual cycle lengths and Hispanic and Black women had slightly shorter cycles than White non-Hispanic women (Li et al., 2023). Further research on the FP and LP variability still needs to be done on more ethnically diverse cohorts.
Conclusion
This physiological investigation of serial menstrual cycles and ovulation over an approximate year describes the menstrual cycle, FP, and LP length variability in a cohort of healthy, premenopausal women with two initially normal-length and normally ovulatory menstrual cycles. We confirm results from previous studies that the FP is more variable than the LP both between-women and within-woman. However, our results highlight the contribution of LP lengths to the variability of the menstrual cycle length and counter the oft-quoted idea that the LP is stable 13–14 days long. In addition, our study contributes to the evidence that SOD are prevalent in normal-length, ‘regular’ menstrual cycles. These disturbances may be part of a continuum of adaptable and reversible reproductive changes in response to physiological and psychosocial stressors and may have implications for women’s reproductive, bone, and other, as yet unstudied, aspects of women’s health.
Data availability
The data underlying this article will be shared upon reasonable request to the corresponding author.
Acknowledgements
The authors thank the participating women in the Prospective Ovulation Cohort (Prior et al., 1990a) for their volunteered time and commitment. We appreciate Yvette Vigna BA, RN, for her effort and dedication in creating the Prospective Ovulation Cohort Database and Dhani Kalidasan, MSc of the Centre for Menstrual Cycle and Ovulation Research (CeMCOR) for her ongoing support in facilitating the research resources needed for completion of this project.
Authors’ roles
J.C.P. designed and collected the original dataset. J.C.P. and S.H. conceived the objectives of this study. S.H. and S.S. performed the statistical analysis and compiled the tables and figures. A.G. was instrumental in selecting the 53-woman cohort for analysis and providing editing. S.H. wrote the manuscript with guidance and feedback from J.C.P., A.G., and S.S.
Funding
Canadian National Health Research Development Project (NHRDP) with supplemental (arm’s length donations) from the Dairy Bureau of Canada. This study was part of a FLEX medical education project of University of British Columbia and funded by donations to CeMCOR supporting the contributions of Kalidasan plus of Dr Goshtasebi and Dr Shirin.
Conflict of interest
The authors do not have any commercial or other conflicts of interest to disclose.
