Menstrual cycle length and adverse pregnancy outcomes among women in Project Viva

Abstract

Background:

Retrospective studies suggest that menstrual cycle length may be a risk marker of adverse pregnancy outcomes, but this evidence is susceptible to recall bias.

Objective:

To evaluate the prospective association between menstrual cycle length and the risk of adverse pregnancy outcomes.

Methods:

Secondary analysis of 2046 women enrolled in Project Viva at ~10 weeks of gestation and followed through delivery. The exposure was menstrual cycle length. The outcomes included gestational glucose tolerance (gestational diabetes/impaired glucose tolerance [GDM/IGT], isolated hyperglycemia), hypertensive disorders of pregnancy (gestational hypertension/preeclampsia), gestational weight gain, birthweight-for-gestational age z-scores (BWZ) categorized in tertiles, preterm birth, and birth outcome (live birth, pregnancy loss). We used modified Poisson and multinomial logistic regression adjusted for age, race/ethnicity, parity, age at menarche, and pre-pregnancy body mass index.

Results:

Mean (SD) age at enrollment was 32.1 (4.9) years. Most women (74.3%) had a cycle length of 26-34 days (reference group), 16.2% reported short cycles (≤25 days), and 9.5% long/irregular cycles (≥35 days/too irregular to estimate). Compared to the reference group, women with short cycles had lower odds of GDM/IGT (odds ratio [OR] 0.50, 95% confidence interval [CI] 0.28, 0.89), whereas women with long/irregular cycles had higher odds (OR 1.72, 95% CI 1.04, 2.83). Additionally, women with short cycles had higher odds of having a newborn in the lowest tertile of BWZ (OR 1.45, 95% CI 1.06, 1.98). There was a U-shaped relation between cycle length and preterm birth with both short (relative risk [RR] 1.49, 95% CI 0.98, 2.27) and long/irregular (RR 2.04, 95% CI 1.30, 3.20) cycles, associated with a higher risk.

Conclusions:

Variation in menstrual cycle length may be a risk marker of GDM/IGT, lower birth size, and preterm birth and flag women who may benefit from targeted monitoring and care before and during pregnancy.

BACKGROUND

Menstrual cycle characteristics are a vital sign of women’s health during the reproductive years.¹ Menstrual dysfunction, manifested as long or irregular menstrual cycles, occurs in 19% of reproductive-aged women.² Long or irregular cycles are of concern as they may be a marker of underlying hormonal imbalances such as insulin resistance and reproductive disorders including infertility and infertility-related gynecologic conditions like endometriosis and polycystic ovary syndrome (PCOS)^3-5 that have implications for short- and long-term health.

Long or irregular menstrual cycles have been associated with an increased risk of coronary heart disease and type 2 diabetes in mid-life, and premature mortality.^6-9 Previous studies, mostly retrospective, have also suggested an association between long or irregular cycles and adverse pregnancy outcomes, including hypertensive disorders of pregnancy (HDP),^10,11 gestational diabetes mellitus (GDM),¹² preterm birth,¹³ and low birthweight.¹¹ Yet, inference from the current literature is limited by biases that plague retrospective and case-control studies, including recall and selection bias.^14,15 Furthermore, results have been inconsistent in part due to varied and non-standard definitions of menstrual dysfunction, including definitions based on cycle length to differentiate short or long cycles,^10,12,16,17 the frequency of cycles,¹² self-reported irregular periods,^13,18 or deviation of >7 days between self-reported and ultrasound derived gestational age.¹¹

In this study, we evaluated the association of clinically relevant thresholds of menstrual cycle length ¹⁹ with the risk of adverse pregnancy outcomes using prospectively collected information from a longitudinal cohort.

METHODS

Study design and cohort

This study was a secondary analysis of data from Project Viva, an ongoing prospective cohort of women recruited between 1999 and 2002 from Atrius Harvard Vanguard Medical Associates at around 10 weeks gestation and followed through delivery. Inclusion criteria were as follows: singleton pregnancy, <22 weeks of gestation at recruitment, ability to answer questions in English, and planning to stay in the study area until delivery. Additional details on recruitment and eligibility have been described elsewhere.²⁰ For 30 women who participated with two different pregnancies, we only considered the first pregnancy enrolled in the cohort. Our sample included 2276 women with a singleton live birth (n=2100) or pregnancy loss (n=176). We excluded women younger than 18 years at enrollment (n=30), those with pregestational chronic hypertension, type 1 or type 2 diabetes (n=45), and women who responded “no periods” or “don’t know” (n=45) to the question regarding their menstrual cycle length, or who had missing information on this question (n=110). Our analytical sample included 2046 women with data on menstrual cycle length and at least one adverse pregnancy outcome (Figure 1).

Figure 1.

Flowchart of the study sample

Exposure

At the first study visit (~10 weeks of gestation), participants reported their menstrual cycle length in response to the question, “When you are not on the pill, breastfeeding or pregnant, what is the typical length of your menstrual cycle? By this, we mean the interval from the first day of your period to the first day of your next period”. Response options included <21 days, 21-25 days, 26-34 days, ≥35 days, or too irregular to estimate. For analysis, we combined the categories of <21 days with 21-25 days and ≥35 days with too irregular to estimate because of small cell sizes for these categories.

Outcomes

We obtained information on the development of adverse pregnancy outcomes by review of outpatient and hospital medical records. Using information from clinical GDM screening at 26-28 weeks of gestation, we classified women as having GDM, impaired glucose tolerance (IGT), isolated hyperglycemia, or normoglycemia.²¹ We combined GDM and IGT due to the small sample in each group.

We used outpatient and hospital medical records to classify women as normotensive, gestational hypertension (GH), and preeclampsia.²² We combined GH and preeclampsia in a single outcome (Hypertensive Disorders of Pregnancy – HDP) due to the small sample in each group.

Using serial clinical prenatal weights, we calculated total gestational weight gain as the difference between the last clinically measured weight (within four weeks before delivery) and self-reported pre-pregnancy weight and categorized it as inadequate, adequate, or excessive.²³

We obtained information on the newborn’s sex, birthweight, and delivery date from medical records. We calculated birthweight-for-gestational age and sex z-scores (BWZ) based on United States national reference data,²⁴ and categorized it in tertiles due to small sample sizes in some cells when categorizing based on conventional categories of birth size. Gestational age at birth (weeks) was calculated by subtracting the date of the last menstrual period from the date of delivery or from the 2^nd-trimester ultrasound in cases where the estimated delivery date by last menstrual period differed by >10 days.²² We categorized newborns as preterm if they were born <37 weeks of gestation. We obtained information on birth outcome status from outpatient and hospital medical records and classified it as live birth or pregnancy loss (stillbirth or miscarriage).

Covariates

Participants reported their age, race/ethnicity, education level, marital status, annual household income, parity, and smoking habits at enrollment. We calculated pre-pregnancy body mass index (BMI, kg/m²) from self-reported pre-pregnancy weight and height. Women provided information on their age at the first menstrual period at a study visit conducted ~13 years after enrollment. We determined history of infertility for the index pregnancy based on self-reported time to pregnancy ≥12 months (or ≥6 months if ≥35 years) at the first prenatal visit or a diagnosis of infertility or claims for infertility treatments from medical records within a few months of enrollment. We identified additional reports of infertility with a detailed reproductive questionnaire completed ~18 years after enrollment. In this questionnaire, participants reported their time to pregnancy (i.e., ≥12 months or ≥6 months if ≥35 years) or the use of medically assisted reproduction for all their pregnancies, including the index.

Statistical analysis

We assessed the distribution of maternal characteristics across categories of menstrual cycle length and compared them using mean (standard deviation [SD]) for continuous variables or frequencies and proportions for categorical variables.

In multivariable analyses, we examined the associations between short (≤25 days), and long/irregular menstrual cycles (≥35 days/too irregular to estimate), compared to cycles with a usual length of 26-34 days (reference), with risk of the various outcomes of interest. We used modified Poisson regression models with robust variance to estimate the relative risk (RR) and 95% confidence interval (CI) for dichotomous outcomes (HDP, preterm birth, and birth outcome).²⁵ For categorical outcomes (gestational glucose tolerance status, gestational weight gain, and tertiles of BWZ), we used multinomial logistic regression models to estimate the odds ratio (OR) and 95% CI.

We constructed a series of models adjusted for potential confounders for each outcome. Model 1 included age at enrollment (18-29, 30-34, ≥35 years), race/ethnicity (white, Black, Asian, Hispanic, other), parity (0, ≥1 prior birth), and age at menarche (<12, 12-14, ≥15 years). Model 2 further adjusted for pre-pregnancy BMI (continuous). We evaluated the presence of interactions between age at enrollment and parity with menstrual cycle length using interaction terms; none of these were significant (p> 0.05), so final models did not include these product terms. For all the outcomes, additional adjustment for education, marital status, household income, smoking, family history of type 2 diabetes (only for glucose tolerance status), and family history of hypertension (only for HDP) did not influence the results substantially. Therefore, we did not include these variables in the final models.

We conducted all the analyses in Stata 16 (StataCorp L.P., College Station, Texas).

Missing data

To reduce bias due to missing values for covariates, we conducted chained equation multiple imputation to generate 50 imputed data sets using an imputation model that included the exposure, outcomes, and covariates under study. Missingness in covariates varied from <1% for race/ethnicity and pre-pregnancy BMI to 50% for age at menarche. The imputed data sets were combined and analyzed using MI ESTIMATE in Stata 16. A complete case analysis comprising women without missing covariates (n=1012) yielded similar results. Thus, the results included herein are based on imputed covariate data for larger sample sizes.

Sensitivity analyses

First, we repeated the analyses after excluding 56 women who reported cycles of <21 days from the short cycles category to assess whether our findings for this group could be driven by women with polymenorrhea. Second, to minimize the possibility of exposure misclassification, we excluded 93 women who reported that their cycles were too irregular to estimate from the long/irregular cycles category. Third, to assess the extent to which PCOS could explain our findings, we excluded 39 women diagnosed with this condition before the index pregnancy. Finally, because causes of infertility may also be linked with irregular cycles and adverse pregnancy outcomes, we further accounted for history of infertility for the index pregnancy (yes/no) in multivariable models and assessed for changes in the direction, magnitude, and precision of results.

ETHICS APPROVAL

All participants provided written informed consent at enrollment. The institutional review board of Harvard Pilgrim Health Care approved all study protocols.

RESULTS

Attrition analysis

The analytical sample comprised 2046 women who, compared to those excluded (n=230), were older at enrollment (mean age 32.1 vs. 29.7 years) and more likely to be non-Hispanic white (68.3 vs. 45.9%), college-educated (65.4 vs. 46.9%), married or cohabiting (92.0 vs. 80.9%), and to have a household income>$70,000/year (61.6 vs. 52.7%). The analytical sample also included a lower proportion of smokers during pregnancy (12.4 vs. 18.5%).

Sample characteristics

Women in this study were of diverse race/ethnicity and had a mean (SD) age of 32.1 (4.9) at enrollment (Table 1). Most participants (74.3%) had a usual cycle length of 26-34 days, with 16.2% and 9.5% reporting short and long/irregular cycles, respectively. Women with short cycles were less frequently non-Hispanic white, college-educated, and high income. The opposite was observed among women with a usual cycle length of 26-34 days or long/irregular cycles. Women with short cycles also had a higher pre-pregnancy BMI (mean 26.1 kg/m², SD 6.5) than women in the reference group (mean 24.6 kg/m², SD 5.3). The distribution of adverse pregnancy outcomes by menstrual cycle length is described in Supplemental eTable 1.

Table 1.

Participants’ characteristics overall and by menstrual cycle length (n=2046) ^a

			Menstrual cycle length
MATERNAL CHARACTERISTICS	Overall n=2046		≤25 days n=332 (16.2%)	26-34 days n=1520 (74.3%)	≥35 days/too irregular to estimate n=194 (9.5%)
	Mean (SD)		Mean (SD)	Mean (SD)	Mean (SD)
Pre-pregnancy BMI, kg/m2	24.9 (5.6)		26.1 (6.5)	24.6 (5.3)	24.9 (6.0)
	N	%	%	%	%
Age at enrollment
18-29 years	604	29.5	36.7	27.0	37.1
30-34 years	863	42.2	38.0	42.7	45.4
≥35 years	579	28.3	25.3	30.3	17.5
Race/ethnicity
White	1389	68.3	47.1	72.2	74.6
Asian	114	5.6	4.5	5.6	7.8
Black	320	15.7	30.8	13.1	10.9
Hispanic	136	6.7	11.8	5.9	4.1
Other	74	3.6	5.7	3.3	2.6
College graduate
No	703	34.6	51.1	31.1	33.7
Yes	1330	65.4	48.9	68.9	66.3
Married/cohabiting
No	163	8.0	15.1	6.1	10.9
Yes	1869	92.0	84.9	93.9	89.1
Household income>$70,000/year
No	691	38.4	50.6	35.9	41.3
Yes	1107	61.6	49.4	64.1	58.7
Pregnancy smoking status
Former smoker	380	19.3	14.1	20.4	19
Smoked during pregnancy	245	12.4	13.4	11.8	15.9
Never smoker	1349	68.3	72.5	67.9	65.1
Primiparae
No	1063	52.0	54.8	52.4	43.3
Yes	983	48.0	45.2	47.6	56.7
Age at first period
<12 years	172	17.0	20.1	17.3	9.4
12 to 14 years	732	72.2	67.2	72.7	75.0
≥15 years	110	10.8	12.7	9.9	15.6

BMI: body mass index.

^aThe sample includes 2046 women with data on the exposure and ≥1 adverse pregnancy outcome. The description was conducted using non-imputed data. The sample may not add to 2046 due to missing covariates values.

Menstrual cycle length and adverse pregnancy outcomes

Compared to women whose usual cycle length was 26-34 days, women with short cycles (≤25 days days) had lower odds of GDM/IGT (OR 0.50, 95% CI 0.28, 0.89), whereas women with long/irregular cycles (≥35 days/too irregular to estimate) had higher odds of GDM/IGT (OR 1.72, 95% CI 1.04, 2.83), even after accounting for pre-pregnancy BMI (Table 2, model 2).

Table 2.

Relative risk (RR) or odds ratio (OR) and 95% confidence interval (CI) of adverse pregnancy outcomes by menstrual cycle length ^a

PREGNANCY OUTCOME	Unadjusted	Model 1	Model 2
	RR (95% CI)	RR (95% CI)	RR (95% CI)
Hypertensive disorders of pregnancy, n=1865 (gestational hypertension/ preeclampsia vs. normotensive)
≤25 days	0.87 (0.59, 1.30)	0.87 (0.58, 1.29)	0.79 (0.52, 1.18)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	1.18 (0.77, 1.82)	1.17 (0.76, 1.81)	1.11 (0.73, 1.71)
Preterm birth, n=1891 (<37 weeks vs. ≥ 37 weeks)
≤25 days	1.72 (1.15, 2.56)	1.52 (1.00, 2.31)	1.49 (0.98, 2.27)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	2.09 (1.34, 3.26)	2.07 (1.32, 3.25)	2.04 (1.30, 3.20)
Birth outcome, n=2046 (pregnancy loss vs. live birth)
≤25 days	1.33 (0.91, 1.94)	1.33 (0.91, 1.94)	1.30 (0.88, 1.90)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	0.93 (0.53, 1.61)	0.98 (0.56, 1.71)	0.95 (0.54, 1.67)
	OR (95% CI)	OR (95% CI)	OR (95% CI)
Gestational glucose tolerance status, n=1854 (reference: normoglycemic)
GDM/IGT
≤25 days	0.65 (0.38, 1.11)	0.57 (0.33, 1.00)	0.50 (0.28, 0.89)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	1.68 (1.05, 2.69)	1.90 (1.16, 3.10)	1.72 (1.04, 2.83)
Isolated hyperglycemia
≤25 days	1.00 (0.65, 1.54)	1.13 (0.72, 1.76)	1.09 (0.70, 1.70)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	0.68 (0.36, 1.29)	0.76 (0.40, 1.45)	0.74 (0.39, 1.41)
Gestational weight gain, n=1846 (reference: adequate)
Inadequate
≤25 days	1.31 (0.86, 2.00)	1.25 (0.80, 1.93)	1.22 (0.79, 1.90)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	1.30 (0.81, 2.11)	1.33 (0.82, 2.17)	1.29 (0.79, 2.10)
Excessive
≤25 days	1.22 (0.90, 1.64)	1.27 (0.94, 1.73)	1.22 (0.89, 1.66)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	0.83 (0.59, 1.19)	0.81 (0.56, 1.16)	0.76 (0.52, 1.09)
Tertiles of BWZ, n=1890 (reference: tertile 2) b
Tertile 1
≤25 days	1.56 (1.16, 2.12)	1.45 (1.06, 1.99)	1.45 (1.06, 1.98)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	1.43 (0.98, 2.09)	1.36 (0.93, 2.00)	1.35 (0.92, 1.99)
Tertile 3
≤25 days	1.05 (0.76, 1.44)	1.13 (0.81, 1.57)	1.09 (0.78, 1.52)
26-34 days	1.00 (Reference)	1.00 (Reference)	1.00 (Reference)
≥35 days/too irregular to estimate	1.07 (0.72, 1.58)	1.09 (0.73, 1.63)	1.05 (0.70, 1.57)

BWZ: birthweight-for-gestational age and sex z-scores; GDM: gestational diabetes mellitus; IGT: impaired glucose tolerance.

^aThe sample includes 2046 women with data on the exposure and ≥1 adverse pregnancy outcome.

^bMean (SD) birthweight-for-gestational age and sex z-scores in tertile 1: −0.85 (0.48); tertile 2, 0.17 (0.24); tertile 3, 1.26 (0.50).

Model 1: age at enrollment (18-29, 30-34, ≥35 years), race/ethnicity (white, Black, Asian, Hispanic, other), parity (0, ≥1), age at menarche (<12, 12 to 14, ≥15 years).

Model 2: model 1 + pre-pregnancy body mass index (continuous).

Additionally, women with short cycles had approximately 1.5 times the odds of having a newborn in the lowest tertile of BWZ than women in the reference group across all models (Table 2). For women with long/irregular cycles, this association was in the same direction but weaker.

Finally, women with both short (RR 1.49, 95% CI 0.98, 2.27) and long/irregular (RR 2.04, 95% CI 1.30, 3.20) cycles had higher risk of preterm birth than those with normal cycle length, although the estimate for the former became slightly attenuated after adjustment for pre-pregnancy BMI (Table 2, model 2).

We noted a suggestive association of short cycles with risk of pregnancy loss, but the 95% CIs were wide. We did not observe consistent associations with any of the other outcomes examined.

Sensitivity analyses

Sensitivity analyses changed the findings for GDM/IGT but not for any other outcome. Although excluding women who reported cycles of <21 days from the analysis did not change the association with GDM/IGT (Supplemental eTable 2), long/irregular cycles were no longer associated with GDM/IGT (OR 0.94, 95% CI 0.41, 2.16) after excluding women who reported that their cycles were too irregular to estimate from the analysis (Supplemental eTable 3), or after excluding women diagnosed with PCOS before the index pregnancy (OR 1.23, 95% CI 0.69, 2.20) (Supplemental eTable 4). Finally, adjustment for history of infertility did not influence our findings (Supplemental eTable 5).

COMMENT

Principal findings

Using information from a prospective pregnancy cohort, we showed that women with long/irregular menstrual cycles (≥35 days/too irregular to estimate) had higher odds of GDM/IGT and higher risk of preterm birth. In contrast, women with short cycles (≤25 days) had lower odds of GDM/IGT, higher odds of having a newborn in the first tertile of BWZ, and higher risk of preterm birth. However, the associations of long/irregular cycles with GDM/IGT were substantially attenuated after excluding women reporting that their cycles were too irregular to estimate, perhaps because of loss of statistical power as this resulted in the exclusion of 71% of women with GDM/IGT, and similarly after exclusion of women with a diagnosis of PCOS preceding pregnancy. Overall, our findings suggest that variation in menstrual cycle length may serve as a marker to identify women at risk of common pregnancy complications like GDM and preterm birth.

Strengths of the study

Strengths of this study include the prospective design, the ascertainment of pregnancy outcomes from medical records, the racially/ethnically diverse study sample, the use of clinically relevant categories of cycle length,¹⁹ and rich covariate data. Further, women reported their usual menstrual cycle characteristics early in pregnancy, before the occurrence of any of the adverse pregnancy outcomes, thereby minimizing recall bias.

Limitations of the data

First, although the validity of menstrual cycle characteristics has been previously documented,^26,27 we do not have the necessary data to do so in this cohort. Therefore, we cannot rule out the possibility of exposure misclassification. However, misclassification would likely be non-differential with respect to outcome status. Furthermore, since we assessed menstrual cycle length before the development of adverse pregnancy outcomes, recall bias, if any, is unlikely influenced by outcomes. Second, oral contraceptives can affect menstrual cycle length, but we did not have information on its use before enrollment in the study. However, most of our participants were planning their index pregnancy; hence, most women in our sample were likely not using contraceptives. Third, we used self-reported pre-pregnancy weight to derive BMI, which may be subject to reporting bias. However, self-reported weights are strongly correlated with clinically-measured weights in Project Viva (r=0.99).²⁸ Fourth, we categorized BWZ in tertiles rather than conventional categories of small (SGA)-, appropriate (AGA)- and large-for-gestational age due to sample size considerations. For similar reasons, we combined GDM with IGT, GH with preeclampsia, and the most extreme categories of short and long menstrual cycles. This may have resulted in a lack of granularity for the distinction between clinically relevant events and characteristics. Furthermore, for the BWZ categories, the first tertile of BWZ comprised 16.7% SGA and 83.8% AGA. Given that SGA and AGA represent groups with distinct health risks, the clinical utility of the lowest BWZ tertile is limited. Fifth, we cannot rule out the possibility of unknown or undiagnosed cases of PCOS or other reproductive disorders such as endometriosis that may account for the observed associations. Finally, given that the characteristics of women included vs. excluded were somewhat different, we cannot rule out the possibility of selection bias, which arises when the inclusion of study participants depends on both the exposure and the outcome.²⁹ In our study, it is possible that exposure status (menstrual cycle length) influenced, to some extent, selection into the analytical sample but this is unlikely to be the case for outcome status (adverse pregnancy outcomes). Considering this, we believe that any bias in our results due to differences in characteristics of participants who were included vs. not included is likely non-differential with respect to the relationships of interest and, therefore, skew the estimates towards the null. In addition, we only excluded ~3% of the cohort participants from the analytical sample, so any selection bias, if present, is likely minimal.

Interpretation

Cycle length and GDM/IGT

We observed lower odds of GDM/IGT among women with short cycles and higher odds in women with long/irregular cycles. Our observations in women with long/irregular cycles align with those of some prior studies. Haver et al. found over threefold increased odds of GDM among women with irregular menstrual cycles, defined as cycles that occurred at intervals ≥60 days, those that did not occur monthly, or those that were too irregular to estimate.¹² In a prospective cohort study, Dishi et al. observed 1.78 higher odds of GDM in women with long cycles (i.e., ≥36 days), compared to cycles of 25-30 days; however, the estimates became attenuated after accounting for pre-pregnancy BMI. Contrary to our findings, the Dishi et al. did not observe an association between short cycles (i.e., <24 days) and GDM.¹⁷ Consistent with our findings, a recent prospective study of ~11000 participants in the Nurses’ Health Study II found that women with long cycles during mid-adulthood (i.e., ≥32 days), vs. those with cycles of 26-31 days, had 42% higher risk of GDM.³⁰

Irregular menstrual cycles – particularly long cycles – are often a manifestation of underlying hormonal imbalances and metabolic conditions, including PCOS.^4,31,32 Menstrual dysfunction is a characteristic feature of PCOS with an estimated prevalence of oligomenorrhea (menstrual cycles ≥35 days) ranging from 75 to 85%.³³ Moreover, PCOS has been associated with an increased risk of adverse pregnancy outcomes, including GDM.³⁴ When we excluded women with a PCOS diagnosis before the index pregnancy in a sensitivity analysis, the association between long/irregular cycles and GDM/IGT was attenuated. These findings suggest that the association of cycle length with GDM/IGT may reflect an association caused by PCOS or underlying insulin resistance associated with PCOS, rather than an effect of physiological variation in cycle length.

The association between short menstrual cycles and lower odds of GDM/IGT was unexpected. This association might be related to residual confounding by women’s characteristics, including lifestyle factors. For example, adherence to the Mediterranean diet has been associated with a lower risk of GDM, and it was recently inversely associated with cycle length.^35,36 Our results suggest that at least as regards the risk of GDM/IGT, short and long cycles may represent distinct entities indicative of different metabolic profiles.

Cycle length and birth size

Women with short cycles had higher odds of having a newborn in the first tertile of BWZ. To the best of our knowledge, only one other study has assessed menstrual cycle characteristics in relation to birth size.¹¹ In this retrospective study by Bonnesen et al., they did not find a difference in the risk of having a newborn SGA among women with menstrual irregularities vs. those with regular cycles.¹¹ The inconsistent findings may be due to differences in the definition of the exposure (i.e., irregular menstrual cycles defined as a deviation between self-reported and ultrasound examination-based gestational age of >7 days vs. short or long/irregular cycles based on length), as well as the different categorization of the outcome (SGA vs. tertiles of BWZ).

It is possible that our finding on the association between short menstrual cycles and birth size is related to the presence of reproductive disorders that induce alterations in the endometrium, myometrium, and placenta and increase the risk of adverse pregnancy outcomes.³⁷ For example, endometriosis and hypothyroidism have been linked to short menstrual cycles and are associated with increased risk of small size at birth and low birth weight,^5,38-40 and could partially account for the observed relations.

Cycle length and preterm birth

We observed a U-shaped association between menstrual cycle length and preterm birth. Consistent with our findings, in a population-based case-control study in Iran, women who reported irregular menstrual cycles had higher odds of preterm birth (OR 1.77, 95% CI 1.14, 3.01) than those with regular cycles.¹³ In another population-based prospective study in Australia, Rowlands et al. showed that irregular menstrual periods (i.e., experienced sometimes or often) were associated with greater odds of preterm birth (OR 1.58, 95% CI 1.10, 2.28) but only among women with ≥1 prior birth.¹⁸ In these two studies, they did not characterize menstrual cycle length, so it is not clear whether irregular cycles referred to short or long cycles.

Reproductive disorders could partly explain the relationship between irregular menstrual cycles and preterm birth. For instance, PCOS has been associated with an increased risk of preterm birth.³⁴ However, excluding women with a PCOS diagnosis before the index pregnancy in a sensitivity analysis did not change the association between long/irregular cycles and preterm birth. Our findings suggest that women without a diagnosis of PCOS but who experience oligomenorrhea or cycles too irregular to estimate may nonetheless have metabolic disturbances such as insulin resistance, hyperandrogenism, and a proinflammatory environment that may increase the risk of preterm birth.³⁷

Endometriosis is another disorder, common in reproductive-aged women,⁴¹ that can be accompanied by both long/irregular or short menstrual cycles^5,40 and has been associated with preterm birth.^5,42 Another such reproductive condition is undiagnosed uterine fibroids which have been associated with short menstrual cycles and an increased risk of preterm birth.^43,44 Future studies in human cohorts, as well as in vitro or in vivo mechanistic studies, are needed to clarify the role of physiologic variation in cycle length vs. the effect of specific pathologies.

CONCLUSIONS

Variation in menstrual cycle length may be a marker of GDM/IGT, lower birth size, and preterm birth risk. These findings indicate the potential value of menstrual cycle length to flag women at increased risk of common pregnancy complications and adverse outcomes. Women with short or long/irregular cycles may benefit from targeted monitoring and care before and during pregnancy.

Social media quote.

Variation in menstrual cycle length may be a risk marker of gestational diabetes/impaired glucose tolerance, lower birth size, and preterm birth. Women with short or long/irregular menstrual cycles may benefit from targeted monitoring and care before and during pregnancy.

Synopsis.

Study question

Is there an association between menstrual cycle length and the risk of adverse pregnancy outcomes?

What is already known

Scant studies, mostly retrospective, suggest an association between menstrual cycle length and adverse pregnancy outcomes. Prospective studies are needed to clarify these associations.

What this study adds

Variation in menstrual cycle length may flag women at higher risk of common pregnancy complications such as gestational diabetes/impaired glucose tolerance, preterm birth, and lower birth size.