Menstrual Cycle Characteristics: Associations With Fertility and Spontaneous Abortion : Epidemiology

Secondary Logo

Journal Logo

Original Article

Menstrual Cycle Characteristics

Associations With Fertility and Spontaneous Abortion

Small, Chanley M.*; Manatunga, Amita K.; Klein, Mitchel; Feigelson, Heather S.§; Dominguez, Celia E.; McChesney, Ruth; Marcus, Michele*

Author Information
Epidemiology 17(1):p 52-60, January 2006. | DOI: 10.1097/01.ede.0000190540.95748.e6



Epidemiologists often use menstrual cycle patterns as indicators of endocrine function in environmental and occupational studies, yet few studies have considered whether menstrual cycle characteristics are associated with fertility or pregnancy outcome.


We prospectively studied 470 women to determine whether cycle length or bleed length were associated with fertility or spontaneous abortion. Women completed daily diaries with information on menstrual bleeding, intercourse, birth control use, and covariates. For each menstrual cycle, women collected at least 2 urine samples, which were assayed for human chorionic gonadotropin to define early pregnancies. Women were followed for 1 year or until the end of a clinical pregnancy.


Cycles with lengths of 30 to 31 days preceded cycles with the highest fecundity. Shorter cycles were less likely to be followed by conception (fecundity ratio [FR] = 0.6; 95% confidence interval [CI] = 0.4–1.0). Compared with 30- to 31-day cycles, conceptions after shorter and longer cycles were more likely to be spontaneously aborted (for shorter cycles, odds ratio [OR] = 3.0 [95% CI = 0.9–9.6] and for longer cycles, OR = 3.0 [0.9–10.6]). Cycles with 5 days of menstrual bleeding had the highest fecundity. Cycles with up to 4 days of bleeding had lower fecundity (for bleed lengths of 4 days, FR = 0.5 [0.3–0.8] and for bleed lengths less than 4 days, FR = 0.6 [0.3–0.9]). Spontaneous abortion was less likely after bleeds greater than 5 days (OR = 0.4 [0.1–1.1]) when compared with 5-day bleeds.


Menstrual cycle characteristics appear to be associated with fertility and spontaneous abortion.


Click on the links below to access all the ArticlePlus for this article.

Please note that ArticlePlus files may launch a viewer application outside of your web browser.

Menstrual cycles reflect a complex series of coordinated processes along the hypothalamic–pituitary–ovarian axis. Within and among women, menstrual cycles vary in length and regularity. Epidemiologic studies have found associations between menstrual cycle characteristics and a variety of host, behavioral, occupational, and environmental factors.1–18 The importance of these associations is dependent on the assumption that menstrual cycle characteristics reflect endocrine function and disease risk.

There is accumulating evidence that lifelong menstrual patterns are associated with the risk of chronic diseases, including breast19–24 and ovarian cancer,25,26 uterine fibroids,27 diabetes,28 and cardiovascular disease.29,30 However, few studies have systematically examined whether menstrual cycle characteristics are related to the most direct measures of reproductive health, fertility, and pregnancy outcome.

In the clinical setting, extreme disruptions of the menstrual cycle such as amenorrhea and chronic anovulation are associated with infertility. It is less clear if and how menstrual cycle patterns in the general population influence fertility. The few studies of menstrual cycle characteristics and fertility have design limitations. Two studies relied on self-reported cycle characteristics and time to pregnancy,2,31 whereas a third study did not adjust for intercourse.32 The only prospective study of cycle length and fertility used a single cycle to define a woman's cycle length.33

Our goal was to determine whether menstrual cycles can act as a surrogate measure of reproductive health. The development and recruitment of what becomes the dominant follicle in 1 cycle occurs in the prior cycle,34 and thus the subsequent bleed likely reflects hormonal and physiological components of the prior cycle. Therefore, we used a prospective study to examine whether cycle length of the prior cycle or bleed length of the current cycle was associated with fertility or risk of spontaneous abortion.


Study Population

The Mount Sinai Study of Women Office Workers was designed to assess the reproductive health of women office workers 40 years of age and younger. Women from 14 companies and government agencies in New York, New Jersey, and Massachusetts were enrolled from 1990 through 1994. To determine eligibility for this prospective study of fertility and spontaneous abortion, 4,640 women completed a self-administered questionnaire. Women were eligible if they had been sexually active in the month before completing the questionnaire while using inconsistent or no birth control (n = 855). Women using oral contraceptives or an intrauterine device were not eligible. We excluded women who had a hysterectomy, had been diagnosed with polycystic ovaries, or were currently infertile (had been trying to conceive for more than 12 months) as well as those whose partners had had vasectomies. The study required the completion of daily diaries and urine collection at least 2 days each cycle. Women were asked to participate for 1 year or until the end of a clinical pregnancy. Five hundred sixty-three women agreed to participate. We excluded from the present analyses 79 women who did not collect any urine and 14 women who were found to be ineligible after completing an entry interview. The final sample size was 470 women. The Institutional Review Board at Mount Sinai School of Medicine, New York, NY, and Emory University, Atlanta, GA, approved the protocols, and the participants gave informed consent.

Assessment of Menstrual Cycle Characteristics and Covariates

Participants were interviewed at entry into the study to obtain data on demographic variables, including age and ethnicity, body mass index (BMI, kg/m2), and reproductive history. We used several questions from the interview to determine an approximate number of cycles a woman had been at risk for pregnancy at study entry. Women who were not using any birth control reported the number of months since they last used birth control; women who sporadically used birth control reported the number of months during the past year in which they had intercourse at least once without birth control. We converted the number of months a woman had been at risk for pregnancy before follow up into a number of cycles based on a woman's average cycle length.

During the prospective study, women completed daily diaries containing information on menstrual bleeding, intercourse, birth control use, smoking, alcohol and caffeine consumption. We determined menstrual cycle length and bleed length from the daily diary. Cycle length was defined as number of days from the first day of bleeding until the day before the next bleeding episode, and bleed length was defined as the number of days of menstrual bleeding. When bleeds occurred within 2 days before the start of a bleeding sequence or within 3 days after a bleeding sequence, they were considered part of that sequence.

Assessment of Time to Pregnancy and Spontaneous Abortion

For ascertainment of pregnancies and early spontaneous abortions, women followed 1 of 2 urine collection protocols. Under 1 protocol, 40 women collected daily first-morning urine samples. Under the other protocol, 430 women collected first-morning urine samples on the first 2 days of menses each cycle. If menses did not begin when expected, these women collected 2 urine specimens 1 week after the expected onset of menses. The expected date was based on a self-report of “usual cycle length” from the entry interview. The participants stored samples in home freezers until field workers collected them. The Core Laboratory at the Irving Center for Clinical Research at Columbia University and the Center for Clinical Research at Mount Sinai School of Medicine assayed samples for human chorionic gonadotropin (hCG), a biomarker of early pregnancy. Extensive quality control and split sample comparisons between the 2 labs ensured comparable results. The immunoradiometric assay detection limit was 0.05 ng/mL.35

Among the daily collectors, pregnancies were defined by 2 consecutive days of hCG greater than 0.25 ng/mL in a 4-day window beginning either on the first day of menses or at the expected onset of menses when menses did not occur. Among women who collected 2 samples each cycle, pregnancies were defined by hCG greater than 0.25 ng/mL for both samples. This criterion was derived by analyzing daily urine samples collected from women with tubal ligations. We evaluated 176 cycles from 40 women with tubal ligations, and none met our criterion for pregnancy.

Time to pregnancy was determined by counting the number of menstrual cycles up to and including a conception cycle. We categorized pregnancy outcomes as live birth, subclinical spontaneous abortion, clinical spontaneous abortion, ectopic pregnancy, molar pregnancy, induced abortion, or unknown. A subclinical spontaneous abortion was defined as a cycle with an elevated hCG (meeting the pregnancy criterion described previously) followed by a cycle with no hCG elevation. All of the clinical pregnancies, including the clinical spontaneous abortions, were confirmed by physician diagnosis.

Statistical Analysis

Menstrual cycle and bleed length were each categorized into quintiles based on the number of pregnancies. The crude fecundity of each cycle-length group was calculated as the proportion of cycles within the group that was followed by a conception cycle. The crude fecundity of each bleed-length group was calculated as the proportion of cycles in which conception occurred within the bleed-length group.

We used discrete survival analysis to determine whether menstrual cycle characteristics were associated with time to pregnancy. The discrete time hazard was defined as the conditional probability that a woman became pregnant in a given menstrual cycle conditional on a pregnancy not occurring in prior cycles. The likelihood for a discrete time hazard rate is equivalent to that for binary regression models.36 We modeled a per-cycle probability of conception (fecundity) using logistic regression. We obtained parameter estimates by maximum likelihood and generated fecundity ratios (FRs), representing the odds of conception in 1 group compared with the odds of conception in the referent group, with 95% confidence intervals (CIs).

The analysis of fecundity was based on the first pregnancy that occurred during follow up for each woman. The models incorporated a general baseline hazard using indicator variables for each cycle number and adjusted both for intercourse in a midcycle window and the number of cycles a woman was at risk for pregnancy before follow up. The midcycle window included 7 days before and 2 days after an estimated day of ovulation. Ovulation was estimated by counting 14 days before the onset of subsequent menses. This takes into account the inherent lifespan of the corpus luteum37 and the relative consistency of luteal phase length.5 When menses did not occur, a start date of expected menses was estimated based on a woman's mean cycle length. Previous studies have shown that pregnancy occurs usually with intercourse in a 6-day window ending with ovulation.38–40 Because our estimate of ovulation was inexact, we created a 10-day midcycle window. The number of days of protected intercourse (intercourse with barrier contraceptives, spermicide, or withdrawal) and unprotected intercourse within the midcycle window were entered as 2 continuous cycle-level covariates. As a comparison, we fit models including only unprotected intercourse. The number of cycles that a woman was at risk for pregnancy before follow up was entered as a continuous woman-level covariate.

Other potential predictors of fertility may cause changes in cycle or bleed length, which in turn alter fecundity. We compared models including and excluding the following covariates. Age, BMI, and race were entered as women-level covariates. The average daily number of cigarettes smoked and weekly number of alcoholic and caffeinated beverages consumed were calculated from the daily diaries. Because the menstrual characteristics we examined relate to conditions in the previous cycle, the average values for smoking, caffeine, and alcohol from the previous cycle were entered as continuous cycle-specific covariates.

Finally, we examined whether cycle length (of the previous cycle) or bleed length (of the current cycle) were associated with the probability of spontaneous abortion. This analysis was limited to women who became pregnant. Ectopic pregnancies, pregnancies ending in induced abortion, and pregnancies with unknown outcomes were excluded. We used logistic regression to model the first pregnancy outcome and compared these results with models including subsequent pregnancies.


Our study consists of 470 women followed for a total of 3,774 menstrual cycles. Individual women were followed for 1 to 16 cycles with the median follow up of 7 cycles. Table 1 shows the demographic characteristics and reproductive history of the study population, as reported in the entry interview. The women were predominantly white, married, and well educated. Women were between 19 and 41 years old with a mean age of 31 years. Participants’ BMI ranged from 16.0 to 49.7 kg/m2 with a mean of 24.1 kg/m2. The majority of women (61%) reported a previous pregnancy. The mean age at menarche was 12.8 years. Twenty-three percent of the women reported that they were “trying to get pregnant.” Forty percent of women reported no birth control use during the month before follow up. Fifty-six percent of women were at risk for pregnancy for not more than 2 cycles before follow up.

Baseline Characteristics of 470 Study Participants, Mount Sinai Study of Women Office Workers, 1990–1994

Table 2 shows characteristics of the study population during follow up. Thirty-nine percent of women reported smoking, and 89% reported drinking alcoholic beverages. During the 10-day midcycle window, 64% reported at least 1 day of unprotected intercourse, and 47% reported at least 1 day of protected intercourse.

Follow-Up Characteristics of 470 Study Participants

The median cycle length during follow up was 29.4 days; the mean ± standard deviation (SD) was 30.9 ± 7.5 days. The median bleed length was 5 days; the mean ± SD was 4.9 ± 1.1 days.

During the follow-up period, 179 women (38%) became pregnant with a total of 207 pregnancies (Table 3). Twenty women experienced more than 1 pregnancy, although analyses of fecundity were limited to the first pregnancy. The per-cycle fecundity was 12% for the first cycle and ranged from 9% to 2% in subsequent cycles. Thirty percent of the pregnancies ended in subclinical or clinical spontaneous abortion. The subclinical losses were detected on average 9.7 days earlier than clinical pregnancies.

Outcomes of 207 Pregnancies

Cycle Length and Fecundity

In the cycle-length analyses, we included 160 pregnancies that had a recorded cycle length preceding the conception cycle. Pregnancy was most likely (7.4%) after cycles of 30 to 31 days (Fig. 1). The discrete survival analyses indicate a similar trend (Table 4). Cycle lengths shorter than 30 days or longer than 31 days were less likely to be followed by conception (FR ranging from 0.51–0.74), adjusting for prior risk and intercourse during the midcycle window (model 1). The lowest fecundity followed cycles of 28- to 29-day lengths (FR = 0.51; 95% CI = 0.30–0.87). The combined fecundity ratio for cycles shorter than 30 days was 0.63 (0.41–0.97). Restricting the analysis to clinical pregnancies (model 3) resulted in a slightly more pronounced relationship between cycle length and fecundity. Both the model of all pregnancies (model 1) and the model including only clinical pregnancies (model 3) suggest that cycle lengths longer than 31 days are associated with lower fecundity (FR = 0.69 [0.42–1.12] and 0.63 [0.38–1.03], respectively). Adjusting for other potential predictors of fecundity did not appreciably change the effect estimates for cycle length (models 2 and 4). Controlling for BMI with a quadratic term or categorically did not change the effect estimate of cycle length on fecundity (data not shown).

Unadjusted fecundity in relation to cycle length of the previous menstrual cycle. Vertical bar represents the proportion of cycles within each cycle length category that preceded a pregnancy. N for each category shown above standard error whiskers.
Adjusted Fecundity Ratios Describing a Per-Cycle Probability of Conception

Bleed Length and Fecundity

In the bleed-length analyses, we included 168 pregnancies with recorded bleed lengths at the beginning of the conception cycle. Pregnancy was most likely (7.4%) in cycles with a 5-day bleed (Fig. 2). The discrete survival analyses show a similar trend (Table 5). Compared with cycles with 5-day bleeds, cycles with bleeds equal to 4 days or less than 4 days had lower fecundity (FR = 0.52 [0.33–0.83] and 0.56 [0.34–0.93], respectively), adjusting for prior risk and intercourse in the midcycle window (model 1). There was the suggestion of an association between long bleed lengths and lower fecundity (FR = 0.87 [0.55–1.38] for 6-day bleeds and 0.60 [0.33–1.06] for bleeds longer than 6 days). Restricting the analysis to clinical pregnancies (model 3) or adjusting for other potential predictors of fecundity (models 2 and 4) did not appreciably change the effect estimates of bleed length on fecundity. Controlling for BMI with a quadratic term or categorically also did not change the effect estimate of bleed length on fecundity (data not shown).

Unadjusted fecundity in relation to bleed length of the current menstrual cycle. Vertical bar represents the proportion of cycles within each bleed length category that resulted in pregnancy. N for each bleed length category shown above standard error whisker.
Adjusted Fecundity Ratios Describing a Per-Cycle Probability of Conception

Menstrual Cycle Characteristics and Spontaneous Abortion

Compared with cycles of 30 to 31 days, conceptions after shorter and longer cycles were more likely to end in spontaneous abortion (OR = 3.0 [0.94–9.6] and 3.0 [0.85–10.6], respectively) (Table 6). Conceptions in cycles with bleeds greater than 5 days were less likely to end in spontaneous abortion (OR = 0.36 [0.12–1.06]) when compared with bleed lengths of 5 days. This effect was stronger after adjusting for age, BMI, race, alcohol and caffeine consumption, and cigarette use (OR = 0.30 [0.09–0.94]). Including more than 1 pregnancy per woman resulted in similar effect estimates for both cycle and bleed length and narrower confidence intervals (data not shown).

Unadjusted and Adjusted Odds Ratio for Pregnancy Loss


This is the first prospective study to examine whether cycle-level menstrual characteristics (previous cycle length and current bleed length) are associated with fertility and spontaneous abortion. We found 30- to 31-day cycles and 5-day bleeds preceded the most fecund cycles. Our findings also suggest that cycles of 30 to 31 days and bleed lengths of 5 or more days are associated with a lower risk of spontaneous abortion.

Several biologic mechanisms may underlie a relationship between cycle length and fertility. Short cycles likely reflect shortened follicular phases because 84% of cycle length variability is due to variation in follicular phase length.5 Among women 20 to 39 years of age, short follicular phases are associated with higher follicular phase estrogen.41 Among aging women, changes in gonadotropin and ovarian hormones, including increased follicular phase estrogen,42 are manifest in shortened follicular phases and decreased cycle lengths.43 During this same time, women experience decreased fertility. Correlations between follicular phase length and fertility have been shown in 2 studies of women undergoing fertility treatments.44,45 The decreased fecundity associated with short cycles in our data may result from oocytes of poor quality and hormonal patterns associated with ovarian aging, which may proceed at a different pace than chronologic aging.

A small proportion of short cycles may reflect shortened luteal phases. Short luteal phase length has been associated with decreased progesterone,41 and low progesterone has been associated with decreased pregnancy rates.46 Windham et al41 found decreased progesterone in the luteal phase to be associated with both short and long follicular phases of the subsequent cycles and suggest that a compromised corpus luteum results in perturbation of gonadotropin secretion. Given that follicular development and recruitment of the dominant follicle occur in the previous cycle,34 the low fecundity subsequent to short cycles may result from poor oocyte quality resulting from this perturbation. Anovulation also may be responsible for the decreased fecundity of both short and long cycles. When compared with cycles of 25 to 32 days, longer and shorter cycles are 10% to 30% more likely to be anovulatory.47

Our results are consistent with the only other prospective study of cycle length and fertility. Kolstad et al33 defined a woman's cycle length by her first observed menstrual cycle. They too found the highest crude fecundity after cycles of 30- to 31-day lengths and lower fecundity after short cycles. In a cross-sectional study using self-reported cycle length and infertility (defined as a history of having sexual intercourse a year or more without getting pregnant), Rowland et al2 found infertility associated with cycles 36 days and longer. Cycles shorter than 24 days were not associated with infertility. In a retrospective study, Jensen et al31 found cycles shorter than 28 days and longer than 31 days were associated with low fecundity. Twenty-eight-day cycles were associated with the highest fecundity. Because information was obtained from women during a routine obstetric visit at 20 weeks gestation, this population does not include women who never became pregnant or those who had spontaneous abortions before 20 weeks gestation. All of the spontaneous abortions in our study occurred before 20 weeks. Women who are currently pregnant represent a unique population, and women who get pregnant quickly may be more likely to recall “normal” cycle lengths of 28 days. Long cycles were associated with spontaneous abortions in a previous study of women with repeated miscarriages.48

There is little information on the underlying hormonal basis for bleed length patterns. We postulate that a short bleed may result from a quick drop in estrogen from the prior cycle caused by deficient follicles producing less estrogen. Low follicular-phase estrogen has been associated with lower fertility.49 Additionally, short bleeds may indicate an insufficient buildup of the uterine lining during the previous cycle. If this tendency is repeated in the subsequent cycle, the endometrium may not be adequate for implantation, resulting in low fecundity or high rates of spontaneous abortion. Long bleed lengths are associated with anovulatory cycles,47 possibly explaining the lower fecundity after bleeds longer than 7 days.

One study has examined the effect of bleed length on fecundity. In a population of rural Bolivian women (approximately half of whom were lactating), menstrual bleeding preceding a conception was longer than bleeding not followed by conception.32 Because the mean bleeding duration in this population (3.5 days) was short relative to our sample, this finding is not inconsistent with a 5-day bleed resulting in the highest fecundity.

Our analyses examine how the hormonal and physiological characteristics of the previous cycle, as reflected in cycle and bleed length, relate to the fecundity of the subsequent cycle. Clinical evidence supports this concept. Recent in vitro fertilization studies found that manipulating the hormonal milieu of the preceding luteal phase could optimize oocyte retrieval in the subsequent cycle. Premenstrual (luteal-phase) treatment with estrogen or a GnRH antagonist reduces the pace of early follicular growth and increases the number of viable oocytes in previously poor responding women.50,51 These data support our hypothesis that the hormonal milieu of the previous cycle affects the development of the dominant follicle in the subsequent cycle.

We estimated the number of cycles a woman was at risk for pregnancy before follow up using several interview questions targeted to groups of women with different birth control practices. The estimation may be less accurate for sporadic birth control users. To determine whether the method of estimation influenced our models, we included an interaction between the birth control group (sporadic vs none) and the prior risk covariate. The interaction was not significant and our main effect estimates were not altered.

For our fecundity analyses, we included a covariate to adjust for a woman's prior risk. An alternate approach is to adjust the actual follow-up time to reflect this prior risk. Reanalysis with this approach resulted in effect estimates similar to those reported here. These results appear in an Appendix (available with the online version of this article).

The median cycle length for our population (29.4 days) was slightly longer than has been reported in other large population studies,3,52,53 but was similar to a study of women planning their first pregnancy.33 Different exclusion criteria can result in dissimilar mean cycle length. Although Waller et al5 reported a shorter mean cycle length in a study with a design similar to ours, they excluded amenorrheic women. When we exclude cycles longer than 60 days, our mean length is more similar to that reported by Waller et al, and the effect estimates from our survival models are unchanged.

The conception rate of women discontinuing birth control to become pregnant is 25% to 30% in the first cycle.54,55 Our population had considerably lower fecundity for several reasons. First, we included both pregnancy planners and nonplanners. Only 23% of participants reported that they were “trying to get pregnant” at the time of the entry interview. This population may better reflect intercourse, birth control use, and fecundity in the general population of the United States where almost 50% of pregnancies are unintended.56 Second, some women entered the study after several months of unprotected intercourse without a pregnancy and likely have lower fertility. Finally, lower fecundity may have resulted from the selection of female office workers for this study. Many fertile women who have young children remain outside the workforce for a time and thus were not eligible for this study. As expected, the frequency of intercourse within the midcycle window was a strong predictor of fecundity. Furthermore, it is interesting to note that the percentage of pregnancies in our sample that ended in spontaneous abortion was similar to that reported by Wilcox et al54 despite differences in the study designs.

We were unable to distinguish spotting or intermenstrual bleeds from true menstrual bleeds with our daily diaries. However, when we excluded cycles shorter than 14 days and 1-day bleeds, our effect estimates were similar. We were also unable to differentiate bleed intensity from bleed length. Because excessive blood flow may occur without prolonged flow, we suggest that future studies collect information on both amount and duration of menses. Our study was limited by lack of information on ovulatory status and an exact day of ovulation (and thus on follicular and luteal phase length). Daily urine collection with hormonal analysis is labor- and cost-intensive. Although such studies provide a gold standard for reproductive epidemiology, our study points to the usefulness of information on cycle and bleed length alone.

Our study design afforded several advantages over previous studies. The prospective follow up, including urine collection, allowed us to diagnose early pregnancies, differentiate long cycles from subclinical spontaneous abortions, and control for intercourse. Furthermore, this is the first study to use multiple observations of cycle and bleed length for each woman to determine how cycle-level menstrual characteristics influence fertility. Specific menstrual characteristics appear to be associated with decreased fecundity and increased risk of spontaneous abortion, suggesting that menstrual cycle characteristics may be useful measures of underlying reproductive states. However, the imprecision of our estimates suggests that large sample sizes may be necessary to see these effects. Menstrual cycles may offer epidemiologists a noninvasive, immediate measure of reproductive health and may be useful for studying a variety of host conditions, occupational and environmental exposures.


We thank the labor organizations, SEIU and 9 to 5, for help in recruiting women; and James Godbold, Philip Landrigan, John O'Connor, and Allen Wilcox for their help in the original study design.


1. Harlow SD, Lin X, Ho MJ. Analysis of menstrual diary data across the reproductive life span applicability of the bipartite model approach and the importance of within-woman variance. J Clin Epidemiol. 2000;53:722–733.
2. Rowland AS, Baird DD, Long S, et al. Influence of medical conditions and lifestyle factors on the menstrual cycle. Epidemiology. 2002;13:668–674.
3. Treloar AE, Boynton RE, Behn BG, et al. Variation of the human menstrual cycle through reproductive life. Int J Fertil. 1970;12:77–126.
4. Harlow SD, Campbell B, Lin X, et al. Ethnic differences in the length of the menstrual cycle during the postmenarcheal period. Am J Epidemiol. 1997;146:572–580.
5. Waller K, Swan SH, Windham GC, et al. Use of urine biomarkers to evaluate menstrual function in healthy premenopausal women. Am J Epidemiol. 1998;147:1071–1080.
6. Harlow SD, Campbell BC. Host factors that influence the duration of menstrual bleeding. Epidemiology. 1994;5:352–355.
7. Sternfeld B, Jacobs MK, Quesenberry CP Jr, et al. Physical activity and menstrual cycle characteristics in two prospective cohorts. Am J Epidemiol. 2002;156:402–409.
8. Kato I, Toniolo P, Koenig KL, et al. Epidemiologic correlates with menstrual cycle length in middle aged women. Eur J Epidemiol. 1999;15:809–814.
9. Hornsby PP, Wilcox AJ, Weinberg CR. Cigarette smoking and disturbance of menstrual function. Epidemiology. 1998;9:193–198.
10. Fenster L, Quale C, Waller K, et al. Caffeine consumption and menstrual function. Am J Epidemiol. 1999;149:550–557.
11. Cooper GS, Sandler DP, Whelan EA, et al. Association of physical and behavioral characteristics with menstrual cycle patterns in women age 29–31 years. Epidemiology. 1996;7:624–628.
12. Windham GC, Waller K, Anderson M, et al. Chlorination by-products in drinking water and menstrual cycle function. Environ Health Perspect. 2003;111:935–941.
13. Mendola P, Buck GM, Sever LE, et al. Consumption of PCB-contaminated freshwater fish and shortened menstrual cycle length. Am J Epidemiol. 1997;146:955–960.
14. Thurston SW, Ryan L, Christiani DC, et al. Petrochemical exposure and menstrual disturbances. Am J Ind Med. 2000;38:555–564.
15. Liu Y, Gold EB, Lasley BL, et al. Factors affecting menstrual cycle characteristics. Am J Epidemiol. 2004;160:131–140.
16. Eskenazi B, Warner M, Mocarelli P, et al. Serum dioxin concentrations and menstrual cycle characteristics. Am J Epidemiol. 2002;156:383–392.
17. Harlow SD, Matanoski GM. The association between weight, physical activity, and stress and variation in the length of the menstrual cycle. Am J Epidemiol. 1991;133:38–49.
18. Windham GC, Elkin EP, Swan SH, et al. Cigarette smoking and effects on menstrual function. Obstet Gynecol. 1999;93:59–65.
19. Garland M, Hunter DJ, Colditz GA, et al. Menstrual cycle characteristics and history of ovulatory infertility in relation to breast cancer risk in a large cohort of US women. Am J Epidemiol. 1998;147:636–643.
20. Gomes AL, Guimaraes MD, Gomes CC, et al. A case–control study of risk factors for breast cancer in Brazil, 1978–1987. Int J Epidemiol. 1995;24:292–299.
21. den Tonkelaar I, de Waard F. Regularity and length of menstrual cycles in women aged 41–46 in relation to breast cancer risk: results from the DOM-project. Breast Cancer Res Treat. 1996;38:253–258.
22. Whelan EA, Sandler DP, Root JL, et al. Menstrual cycle patterns and risk of breast cancer. Am J Epidemiol. 1994;140:1081–1090.
23. Zhu K, Beiler J, Hunter S, et al. The relationship between menstrual factors and breast cancer according to estrogen receptor status of tumor: a case–control study in African-American women. Ethn Dis. 2002;12:23–29.
24. Michels-Blanck H, Byers T, Mokdad AH, et al. Menstrual patterns and breast cancer mortality in a large US cohort. Epidemiology. 1996;7:543–546.
25. Parazzini F, La Vecchia C, Negri E, et al. Menstrual factors and the risk of epithelial ovarian cancer. J Clin Epidemiol. 1989;42:443–448.
26. Tavani A, Ricci E, La Vecchia C, et al. Influence of menstrual and reproductive factors on ovarian cancer risk in women with and without family history of breast or ovarian cancer. Int J Epidemiol. 2000;29:799–802.
27. Chen CR, Buck GM, Courey NG, et al. Risk factors for uterine fibroids among women undergoing tubal sterilization. Am J Epidemiol. 2001;153:20–26.
28. Solomon CG, Hu FB, Dunaif A, et al. Long or highly irregular menstrual cycles as a marker for risk of type 2 diabetes mellitus. JAMA. 2001;286:2421–2426.
29. La Vecchia C, Decarli A, Franceschi S, et al. Menstrual and reproductive factors and the risk of myocardial infarction in women under fifty-five years of age. Am J Obstet Gynecol. 1987;157:1108–1112.
30. Solomon CG, Hu FB, Dunaif A, et al. Menstrual cycle irregularity and risk for future cardiovascular disease. J Clin Endocrinol Metab. 2002;87:2013–2017.
31. Jensen TK, Scheike T, Keiding N, et al. Fecundability in relation to body mass and menstrual cycle patterns. Epidemiology. 1999;10:422–428.
32. Vitzthum VJ, Spielvogel H, Caceres E, et al. Vaginal bleeding patterns among rural highland Bolivian women: relationship to fecundity and fetal loss. Contraception. 2001;64:319–325.
33. Kolstad HA, Bonde JP, Hjollund NH, et al. Menstrual cycle pattern and fertility: a prospective follow-up study of pregnancy and early embryonal loss in 295 couples who were planning their first pregnancy. Fertil Steril. 1999;71:490–496.
34. Gougeon A. Ovarian follicular growth in humans: ovarian ageing and population of growing follicles. Maturitas. 1998;30:137–142.
35. O'Connor J, Schlatterer J, Birken S, et al. Development of highly sensitive immunoassays to measure human chorionic gonadotropin, its beta-subunit, and beta core fragment in the urine: application to malignancies. Cancer Res. 1988;48:1361–1366.
36. Collett D. Modeling Survival Data in Medical Research, 2nd ed. Boca Raton: Chapman & Hall; 2003.
37. Yen SSC, Jaffe RB. Reproductive Endocrinology: Physiology, Pathophysiology, and Clinical Management, 4th ed. Philadelphia: WB Saunders; 1998.
38. Dunson DB, Baird DD, Wilcox AJ, et al. Day-specific probabilities of clinical pregnancy based on two studies with imperfect measures of ovulation. Hum Reprod. 1999;14:1835–1839.
39. Dunson DB, Colombo B, Baird DD. Changes with age in the level and duration of fertility in the menstrual cycle. Hum Reprod. 2002;17:1399–1403.
40. Wilcox AJ, Weinberg CR, Baird DD. Timing of sexual intercourse in relation to ovulation. Effects on the probability of conception, survival of the pregnancy, and sex of the baby. N Engl J Med. 1995;333:1517–1521.
41. Windham GC, Elkin E, Fenster L, et al. Ovarian hormones in premenopausal women: variation by demographic, reproductive and menstrual cycle characteristics. Epidemiology. 2002;13:675–684.
42. Prior JC. Perimenopause: the complex endocrinology of the menopausal transition. Endocr Rev. 1998;19:397–428.
43. Klein NA, Battaglia DE, Fujimoto VY, et al. Reproductive aging: accelerated ovarian follicular development associated with a monotropic follicle-stimulating hormone rise in normal older women. J Clin Endocrinol Metab. 1996;81:1038–1045.
44. Fukuda M, Fukuda K, Andersen CY, et al. Characteristics of human ovulation in natural cycles correlated with age and achievement of pregnancy. Hum Reprod. 2001;16:2501–2507.
45. Liss JR, Check JH, Shucoski K, et al. Effect of short follicular phase on conception outcome. Fertil Steril. 2002;77:S17.
46. Wuttke W, Pitzel L, Seidlova-Wuttke D, et al. LH pulses and the corpus luteum: the luteal phase deficiency. Vitam Horm. 2001;63:131–158.
47. Harlow SD. Menstruation and menstrual disorders: the epidemiology of menstruation and menstrual dysfunction. In: Goldman MB, Hatch MC, eds. Women and Health. San Diego: Academic Press; 2000:99–113.
48. Quenby SM, Farquharson RG. Predicting recurring miscarriage: what is important? Obstet Gynecol. 1993;82:132–138.
49. Lipson SF, Ellison PT. Comparison of salivary steroid profiles in naturally occurring conception and non-conception cycles. Hum Reprod. 1996;11:2090–2096.
50. Fanchin R, Castelo-Branco A, Kadoch IJ, et al. Premenstrual administration of gonadotropin-releasing hormone antagonist coordinates early antral follicle sizes and sets up the basis for an innovative concept of controlled ovarian hyperstimulation. Fertil Steril. 2004;81:1554–1559.
51. Fanchin R, Salomon L, Castelo-Branco A, et al. Luteal estradiol pre-treatment coordinates follicular growth during controlled ovarian hyperstimulation with GnRH antagonists. Hum Reprod. 2003;18:2698–2703.
52. Vollman RF. The Menstrual Cycle. Philadelphia: WB Saunders; 1977.
53. Chiazze L Jr, Brayer FT, Macisco JJ Jr, et al. The length and variability of the human menstrual cycle. JAMA. 1968;203:377–380.
54. Wilcox AJ, Weinberg CR, O'Connor JF, et al. Incidence of early loss of pregnancy. N Engl J Med. 1988;319:189–194.
55. Zinaman MJ, Clegg ED, Brown CC, et al. Estimates of human fertility and pregnancy loss. Fertil Steril. 1996;65:503–509.
56. Henshaw SK. Unintended pregnancy in the United States. Fam Plann Perspect. 1998;30:24–29.

Supplemental Digital Content

© 2006 Lippincott Williams & Wilkins, Inc.