Context Although oligomenorrhea has been associated cross-sectionally with insulin
resistance and glucose intolerance, it is not known whether oligomenorrhea
is a marker for increased future risk of type 2 diabetes mellitus (DM).
Objective To prospectively assess risk of type 2 DM in women with a history of
long or highly irregular menstrual cycles.
Design and Setting The Nurses' Health Study II, a prospective observational cohort study.
Participants A total of 101 073 women who had no prior history of DM and who
reported their usual menstrual cycle pattern at age 18 to 22 years on the
baseline (1989) questionnaire.
Main Outcome Measure Incident reports of DM, with follow-up through 1997, compared among
women categorized by menstrual cycle length (5 categories).
Results During 564 333 person-years of follow-up, there were 507 cases
of type 2 DM. Compared with women with a usual cycle length of 26 to 31 days
(referent category) at age 18 to 22 years, the relative risk (RR) of type
2 DM among women with a menstrual cycle length that was 40 days or more or
was too irregular to estimate was 2.08 (95% confidence interval [CI], 1.62-2.66),
adjusting for body mass index at age 18 years and several other potential
confounding variables. The RR of type 2 DM associated with long or highly
irregular menstrual cycles was greater in obese women, but was also increased
in nonobese women (at body mass indexes at age 18 years of <25, 25-29,
and ≥30 kg/m2, RRs were 1.67 [95% CI, 1.14-2.45],
1.74 [95% CI, 1.07-2.82], and 3.86 [95% CI, 2.33-6.38], respectively).
Conclusion Women with long or highly irregular menstrual cycles have a significantly
increased risk for developing type 2 DM that is not completely explained by
obesity.
Long or highly irregular menstrual cycles may be associated with insulin
resistance.1,2 Women with polycystic
ovary syndrome (PCOS), a condition characterized by oligomenorrhea and androgen
excess, have a high prevalence of glucose intolerance and type 2 diabetes
mellitus (DM) that is not simply attributable to being overweight.3,4 Even without documented androgen excess,
oligomenorrhea may be a marker for DM risk. A cross-sectional study involving
Pima Indian women revealed an increased frequency of type 2 DM among those
with a history of very long menstrual cycles.5
Whether menstrual cycle irregularity predicts future risk of type 2
DM remains unknown. The degree to which concomitant obesity confounds the
association between oligomenorrhea and type 2 DM and whether nonobese oligomenorrheic
women have an increased risk for DM remain unclear.
Therefore, we studied women in the Nurses' Health Study (NHS) II to
assess associations between menstrual cycle pattern and risk for subsequent
type 2 DM.
The NHS II is a prospective study of health outcomes among 116 671
female nurses age 24 to 43 years at study inception in 1989. Ninety-two percent
of the participants were white. Participants completed biennial questionnaires
on lifestyle factors and disease. Self-reported baseline information included
age, weight at age 18 years and in 1989, height, race, family history of DM,
menopausal status, history of oral contraceptive use (never, past, current;
duration of use), smoking habits (never, past, current; number of cigarettes
smoked per day), physical activity level, and personal history of DM and hypertension
and associated treatment. Physical activity level was derived from estimation
of the usual duration of participation each week in 8 common activities (walking,
running, jogging, bicycling, calisthenics/aerobics, tennis, lap swimming,
and other aerobic recreation). Information was also collected on usual menstrual
cycle length at age 18 to 22 years (categorized as <21 days, 21-25 days,
26-31 days, 32-39 days, 40-50 days, >50 days, or too irregular to estimate)
and usual cycle pattern at age 18 to 22 years, excluding time around pregnancies
or when using oral contraceptives (very regular [within 3 days], regular,
usually irregular, always irregular, no periods). At baseline, women were
asked about a history of severe teenage acne and, in 1991, about a history
of physician-diagnosed hirsutism. In 1993 and 1997, women were also asked
about a history of infertility treatment with agents to induce ovulation.
Weight was reassessed on each biennial questionnaire.
Women subsequently reporting a new diagnosis of DM were sent supplementary
questionnaires asking about diagnosis, treatment, and history of ketoacidosis,
to confirm the self-report and to distinguish type 1 from type 2 DM.
For this analysis, women were excluded for the following reasons: (1)
did not respond to the question about menstrual cycle length, (2) had DM or
were menopausal at baseline, and (3) diagnosis could not be confirmed by the
supplementary questionnaire to be consistent with type 2 DM. We also excluded
women who reported consistent oral contraceptive use during the years of self-described
cycle characteristics (at least 10 months per year for each year between age
18 and 22 years).
The distribution of person-time by category of usual menstrual cycle
length at age 18 to 22 years was as follows: less than 21 days, 1.0%; 21 to
25 days, 10.0%; 26 to 31 days, 66.1%; 32 to 39 days, 15.3%; and 40 days or
more or too irregular to estimate, 7.6%.
Several studies have confirmed a high accuracy of self-report among
women participating in this study or the NHS, a companion cohort comprising
older nurses. Among a sample of 184 participants in the NHS, the correlation
between reported and measured body weight was 0.96.6
Similarly, among a subset of NHS II participants, correlations between self-reported
weight and height at age 18 years, and weight and height recorded on entry
to nursing school, were 0.87 and 0.94, respectively.7
The validity of supplementary questionnaires to confirm and characterize
DM type has also been validated in the NHS.8
Of a subset of 84 women whose supplementary questionnaire information was
diagnostic of definite type 2 DM, 71 gave permission for medical record review;
records were obtained for 62 of these women. On review by an endocrinologist
blinded to the supplementary questionnaire information, the diagnosis of type
2 DM was confirmed by National Diabetes Data Group criteria in 61 (98%) of
the 62 women.9
We calculated person-time for each participant from the date of return
of the baseline questionnaire to the date of diagnosis of DM or June 1, 1997,
whichever came first. We considered women reporting a usual menstrual cycle
length of 26 to 31 days at age 18 to 22 years to be the referent group, and
women reporting menstrual cycle lengths of 40 days or longer were considered
oligomenorrheic. Relative risks (RRs) and 95% confidence intervals (CIs) for
development of DM were estimated using pooled logistic regression, which approximates
a Cox regression analysis. Covariates obtained from the baseline or subsequent
questionnaires were used in multivariate analyses, including age (continuous),
body mass index (BMI) at age 18 years or currently (6 categories), family
history of DM in a first-degree relative (yes or no), physical activity level
(5 categories of total MET [metabolic equivalent] expenditure), cigarette
smoking (current, past, never), weight change (from BMI at age 18 years to
current BMI), race, and use of oral contraceptives. These covariates were
chosen based on recognized or potential associations between these factors
and risk of DM. We also tested for effect modification by BMI and family history
of DM by performing analyses stratified by these variables. Statistical analysis
was performed using SAS statistical software (SAS Institute Inc, Cary, NC).
Baseline characteristics of the study cohort as a function of menstrual
cycle length are shown in Table 1.
Compared with women with a normal menstrual cycle length (26-31 days), women
with short cycles (<21 days) appeared more likely to be nonwhite, to have
a family history of DM in a first-degree relative, to be current smokers,
and to have a history of oral contraceptive use at study entry in 1989; women
with long (≥40 days) or highly irregular cycles had higher BMI both at
age 18 years and at study entry and greater weight gain from the age of 18
years. Women with either short or long cycles were more likely than women
with a cycle length of 26 to 31 days to report severe teenage acne, physician-diagnosed
hirsutism, ovulatory infertility, and a history of gestational DM; histories
of hirsutism, ovulatory infertility, and gestational DM were particularly
prevalent in the women with long or highly irregular cycles.
During 564 333 person-years of follow-up, we documented 507 cases
of definite type 2 DM. Compared with women whose menstrual cycle length at
age 18 to 22 years was 26 to 31 days, the RR for type 2 DM was modestly but
not significantly elevated among women with a cycle length of less than 21
days both in age-adjusted analyses and after adjustment for BMI at age 18
years, physical activity level in 1989, smoking status in 1989, family history
of DM in a first-degree relative, and history of oral contraceptive use (multivariate
RR, 1.50; 95% CI, 0.70-3.19). Women with usual menstrual cycle lengths of
21 to 25 days or 32 to 39 days had subsequent risks for DM very comparable
to women with a cycle length of 26 to 31 days (Table 2).
In contrast, women with a usual cycle length of 40 days or more or too
irregular to estimate had a significantly increased risk for type 2 DM (multivariate
RR, 2.08; 95% CI, 1.62-2.66) (Table 2).
The Hosmer-Lemeshow goodness-of-fit statistic10
for the main multivariate model with the outcome of type 2 DM was 10.1 (P = .26), indicating a good model fit. Additional adjustment
for updated weight change from age 18 years did not materially affect results
(RR associated with cycle length of ≥40 days or too irregular to estimate,
2.31; 95% CI, 1.77-3.02).
To assess whether the presence of either of 2 markers of androgen excess
modified the association between oligomenorrhea and type 2 DM, we performed
analyses stratified by the presence or absence of a report of physician-diagnosed
hirsutism and/or severe teenage acne. Among the cohort, 6.9% of women reported
severe teenage acne and 2.4% reported physician-diagnosed hirsutism, while
91% of women denied either of these. Even in the absence of these markers
of clinical androgen excess, oligomenorrhea remained a significant predictor
of type 2 DM (multivariate RR, 2.11; 95% CI, 1.59-2.80), while short cycles
were not predictive of DM risk. Among women who reported these markers of
androgen excess, there was a significant association between short cycles
and risk of type 2 DM; long cycles were not a significant predictor of DM
risk, but there were few DM cases in this group (Table 3).
We also assessed the effects of BMI and family history on the relationship
between cycle length and risk of DM (Table
3). Crude rates of DM were 51 per 100 000 person-years among
lean women (BMI, <25 kg/m2) with a cycle length of 26 to 31
days and 84 per 100 000 person-years among lean women with oligomenorrhea;
for those with a BMI of 25 kg/m2 or more, rates were 328 per 100 000
person-years among women with normal cycles and 839 per 100 000 person-years
among women with oligomenorrhea. Associations between oligomenorrhea and type
2 DM appeared slightly stronger among women whose BMI at age 18 years was
greater than 30 kg/m2 than among less obese women. However, even
among women with a BMI of less than 25 kg/m2, a history of long
or highly irregular cycles was a significant predictor of subsequent type
2 DM. Short cycles were associated with an increased risk for DM in women
with a family history of DM in a first-degree relative, but not in women without
this history. However, a history of long or highly irregular cycles was associated
with an increased risk for DM regardless of family history of DM (Table 3).
To minimize the likelihood of screening bias among women with irregular
menstrual cycles, we did an analysis including as cases only the 395 women
reporting symptoms at diagnosis of DM. Compared with women with normal menstrual
cycle length (26-31 days), RRs for symptomatic DM associated with long cycles
remained statistically significant and comparable to those in the primary
analyses: RR, 1.82 (95% CI, 1.35-2.44) in multivariate analyses adjusting
for BMI at age 18 years, physical activity level, smoking, family history
of DM in a first-degree relative, and oral contraceptive use. Short cycles
(<21 days) were not associated with a significant increase in DM risk (multivariate
RR, 1.36; 95% CI, 0.56-3.33). In addition, 89.4% of the cohort reported a
routine physician visit in the previous 2 years, and results were materially
unchanged in analyses limited to these women.
A history of oral contraceptive use of at least 2 months' duration was
highly prevalent at baseline (1989) regardless of cycle length (Table 1). To minimize the possibility that oral contraceptive use
in women with oligomenorrhea might modify risk for DM, we conducted an analysis
excluding women reporting any history of oral contraceptive use of 2 months
or more prior to study entry in 1989. Among this small subgroup (n = 107 cases),
long or highly irregular menstrual cycles remained a significant predictor
of type 2 DM (data not shown); the small number of cases with a cycle length
of less than 21 days precluded valid assessment of the RR associated with
short cycles. We also conducted an analysis updating for oral contraceptive
use through the study period, and results remained essentially unchanged.
The association between long or highly irregular cycles and subsequent
diagnosis of type 2 DM was materially unchanged in additional analyses adjusting
for race or for a history of hypertension and antihypertensive therapy. Furthermore,
results were comparable in an analysis excluding women who reported a history
of infertility treatment (data not shown).
When menstrual cycle patterns at age 18 to 22 years were assessed by
the question on cycle regularity, associations between menstrual cycle abnormalities
and type 2 DM persisted. Compared with women reporting cycles that were usually
or always regular, women reporting cycles that were usually or always irregular
or no cycles had a multivariate RR for type 2 DM of 1.52 (95% CI, 1.26-1.83).
Among this large cohort of women, oligomenorrhea at the age of 18 to
22 years was a significant predictor of subsequent development of type 2 DM.
This observation was not explained simply by associated obesity or several
other potential confounding variables.
In a cross-sectional study, Roumain et al5
reported an increased likelihood of type 2 DM among Pima Indian women with
a menstrual cycle length of 3 months or more, although the risk increase was
not statistically significant after adjustment for obesity. In analyses stratified
by BMI, the association between oligomenorrhea and DM was most pronounced
among the least obese women, ie, BMI less than 30 kg/m2, whereas
rates of DM were high among more obese women regardless of cycle characteristics.
Among our population, the average BMI was much lower than among the
Pima women, as were the rates of type 2 DM. Using a less extreme definition
of oligomenorrhea, we found significant associations between oligomenorrhea
and type 2 DM among women with high BMI (≥30 kg/m2) as well
as lower BMI at age 18 years.
A high rate of glucose intolerance is well-recognized among women with
PCOS,4 a condition typically defined by the
presence of both oligomenorrhea and androgen excess. Studies involving oral
glucose tolerance testing among women with PCOS11,12
demonstrated impaired glucose tolerance in 31% to 35% and DM in 7.5% to 10%.
Abnormalities of glucose homeostasis are greatest in obese women with PCOS
but are also seen in nonobese women with this condition.
Insulin resistance is characteristic of PCOS and appears to underlie
these observations. Women with PCOS have higher fasting and postprandial insulin
levels, and their acute insulin responses to a glucose load are inappropriately
low for the magnitude of peripheral insulin resistance.13
Insulin resistance and associated β-cell dysfunction appear to predispose
to type 2 DM, and high insulin levels have predicted progression to type 2
DM among high-risk populations.14 Concomitant
hyperandrogenism, when present, may have a direct adverse effect on insulin
resistance15 and is also associated with reduced
levels of sex hormone-binding globulin, a marker for reduced insulin sensitivity16 that predicts risk for type 2 DM in women.17
While PCOS is classically defined by both anovulation and hyperandrogenism,
oligomenorrhea in the absence of obvious androgen excess may also frequently
be attributable to PCOS. Studies have reported elevated luteinizing hormone
levels in 90% of oligomenorrheic women18 and
polycystic ovaries on ultrasonography in 87%.19
Elevated testosterone levels have likewise been reported in women with oligomenorrhea
in the absence of clinical hyperandrogenism.20
Greater insulin resistance and higher androgen levels have been noted among
Pima Indian women with long menstrual cycles.1
Menstrual irregularities other than long cycles may also be consistent
with PCOS.21 We found that risk for type 2
DM was significantly higher in women who reported "irregular cycles" irrespective
of cycle length. Risk of DM also appeared higher in women who reported a cycle
length of less than 21 days, although there were relatively few women in this
category and findings were not statistically significant.
While long or irregular cycles are likely to be a marker of PCOS, other
possible causes of oligomenorrhea must be considered. Estrogen deficiency
consistent with hypogonadotropic hypogonadism is observed in a minority of
women presenting with oligomenorrhea or amenorrhea.18
This condition is commonly associated with weight loss, excessive exercise,
or low body weight, factors that would be expected to have a protective effect
on development of DM.22 Inclusion in the current
study of oligomenorrheic women with this condition, or with other conditions
unassociated with DM, would lead to underestimation of the risk associated
with PCOS.
Importantly, absolute rates of DM in this cohort were relatively low
overall, reflecting the relatively young age of the women who, after 8 years
of follow-up, ranged from 32 to 51 years old. The presence of obesity in addition
to oligomenorrhea conferred much greater absolute risk for DM than did oligomenorrhea
alone, consistent with previous observations that obesity is a major modifiable
risk factor for DM.22 Previous studies in women
with documented PCOS have indicated that weight loss reduces insulin resistance
and hyperandrogenism,23 and that leaner women
have lower rates of DM.11
We found that oligomenorrhea was associated with subsequent type 2 DM
even among women who did not report severe teenage acne or hirsutism. However,
we were not able to assess comprehensively the prevalence of clinical androgen
excess and did not have biochemical measures of hyperandrogenism.
One third or more of DM cases are undiagnosed in the general population.24 While screening bias, ie, greater screening for DM
in the women with irregular cycles, might contribute to the observed results,
we consider this an unlikely explanation. It is possible that women with cycle
irregularity may have seen a physician more frequently, but most women in
this cohort of health professionals received routine health care, and the
association between oligomenorrhea and DM was unchanged in analyses limited
to those who had recently seen a physician. Also, until this year,25 PCOS was not among the conditions for which the American
Diabetes Association recommended DM screening, and such screening is not recommended
on the basis of menstrual irregularity alone. Furthermore, results of analyses
limited to women presenting with symptomatic DM, in whom screening bias would
be unlikely to explain results, yielded comparable results. Diagnostic criteria
for DM changed after the time women in this cohort were diagnosed,26 so that some women classified as nondiabetic would
now be considered cases; however, this would not affect the validity of the
findings.
Because oral contraceptives may adversely affect insulin sensitivity
and glucose tolerance,27 a potential concern
is that greater oral contraceptive use among women with irregular cycles might
underlie increased DM risk among these women. However, neither adjustment
for duration of oral contraceptive use, nor exclusion of women reporting a
history of oral contraceptive use, materially changed the results. In addition,
a previous NHS analysis did not find an appreciable increase in DM risk with
prior use of oral contraceptives.28
A limitation of this study is that cycle characteristics are self-reported
retrospectively. Nevertheless, the observed frequency of long or highly irregular
cycles was comparable to that among college students studied cross-sectionally.29 Women whose long or irregular menstrual cycles were
normalized by treatment with oral contraceptives might have classified their
cycles as normal; however, such misclassification would tend to bias toward
the null. Duration of menstrual periods was not assessed.
This study indicates that long or highly irregular menstrual cycles
are a predictor of increased risk for type 2 DM, and that this risk is further
increased by, but not completely explained by, obesity. These findings are
consistent with the suggestion of previous cross-sectional studies that menstrual
cycle irregularities may be a marker for associated metabolic abnormalities
and suggest that women with this history might particularly benefit from lifestyle
approaches to reduce risk, such as weight control and exercise.
1.Weiss DJ, Charles MA, Dunaif A.
et al. Hyperinsulinemia is associated with menstrual irregularity and altered
serum androgens in Pima Indian women.
Metabolism.1994;43:803-807.Google Scholar 2.Dunaif A, Segal KR, Futterweit W, Dobrjansky A. Profound insulin resistance, independent of obesity, in polycystic
ovary syndrome.
Diabetes.1989;38:1165-1174.Google Scholar 3.Chang RJ, Nakamura RM, Judd HL, Kaplan SA. Insulin resistance in nonobese patients with polycystic ovarian disease.
J Clin Endocrinol Metab.1983;57:356-359.Google Scholar 4.Dunaif A, Graf M, Mandeli J, Laumas V, Dobrjansky A. Characterization of groups of hyperandrogenic women with acanthosis
nigricans, impaired glucose tolerance, and/or hyperinsulinemia.
J Clin Endocrinol Metab.1987;65:499-507.Google Scholar 5.Roumain J, Charles MA, De Courten MP.
et al. The relationship of menstrual irregularity to type 2 diabetes in Pima
Indian women.
Diabetes Care.1998;21:346-349.Google Scholar 6.Willett W, Stampfer M, Bain C, Rosner B, Speizer FE. Cigarette smoking, relative weight, and menopause.
Am J Epidemiol.1983;117:651-658.Google Scholar 7.Troy LM, Hunter DJ, Manson JE, Colditz GA, Stampfer MJ, Willett WC. The validity of recalled weight among younger women.
Int J Obes Relat Metab Disord.1995;19:570-572.Google Scholar 8.Manson JE, Colditz GA, Stampfer MJ.
et al. A prospective study of maturity onset diabetes mellitus and risk of
coronary heart disease and stroke in women.
Arch Intern Med.1991;151:1141-1147.Google Scholar 9.National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories
of glucose intolerance.
Diabetes.1979;28:1039-1057.Google Scholar 10.Hosmer DW, Hosmer T, Le Cessie S, Lemeshow S. A comparison of goodness-of-fit tests for the logistic regression model.
Stat Med.1997;16:965-980.Google Scholar 11.Legro RS, Kunselman AR, Dodson WC, Dunaif A. Prevalence and predictors of risk for Type 2 diabetes mellitus and
impaired glucose tolerance in polycystic ovary syndrome: a prospective, controlled
study in 254 affected women.
J Clin Endocrinol Metab.1999;84:165-169.Google Scholar 12.Ehrmann DA, Barnes RB, Rosenfield RL.
et al. Prevalence of impaired glucose tolerance and diabetes in women with
polycystic ovary syndrome.
Diabetes Care.1999;22:141-146.Google Scholar 13.Dunaif A, Finegood DT. B-cell dysfunction independent of obesity and glucose intolerance in
the polycystic ovary syndrome.
J Clin Endocrinol Metab.1996;81:942-947.Google Scholar 14.Lillioja S, Mott DM, Spraul M.
et al. Insulin resistance and insulin secretory dysfunction as precursors
of non-insulin-dependent diabetes mellitus: prospective studies of Pima Indians.
N Engl J Med.1993;329:1988-1992.Google Scholar 15.Polderman KH, Gooren LJ, Asscheman H, Bakker A, Heine RJ. Induction of insulin resistance by androgens and estrogens.
J Clin Endocrinol Metab.1994;79:265-271.Google Scholar 16.Haffner SM, Katz MS, Stern MP, Dunn JF. The relationship of sex hormones to hyperinsulinemia and hyperglycemia.
Metabolism.1988;37:683-688.Google Scholar 17.Haffner SM, Valdez RA, Morlaes PA, Hazuda HP, Stern MP. Decreased sex hormone-binding globulin predicts non-insulin-dependent
diabetes mellitus in women but not in men.
J Clin Endocrinol Metab.1993;77:56-60.Google Scholar 18.Hull MGR. Epidemiology of infertility and polycystic ovarian disease: endocrinological
and demographic studies.
Gynecol Endocrinol.1987;1:235-245.Google Scholar 19.Adams J, Polson DW, Franks S. Prevalence of polycystic ovaries in women with anovulation and idiopathic
hirsutism.
Br Med J (Clin Res Ed).1986;293:355-359.Google Scholar 20.Diamanti-Kandarakis E, Kouli CR, Bergiele AT.
et al. A survey of the polycystic ovary syndrome in the Greek Island of Lesbos:
hormonal and metabolic profile.
J Clin Endocrinol Metab.1999;84:4006-4011.Google Scholar 21.Chang RJ, Katz SE. Diagnosis of polycystic ovary syndrome.
Endocrinol Metab Clin North Am.1999;28:397-408.Google Scholar 22.Hu FB, Manson JE, Stampfer MJ.
et al. Diet, lifestyle, and the risk of type 2 diabetes mellitus in women.
N Engl J Med.2001;345:790-797.Google Scholar 23.Legro RS. Polycystic ovary syndrome: current and future treatment paradigms.
Am J Obstet Gynecol.1998;179:S101-S108.Google Scholar 24.Harris MI, Flegal KM, Cowie CC.
et al. Prevalence of Diabetes, Impaired Fasting Glucose, and Impaired Glucose
Tolerance in US Adults: The Third National Health and Nutrition Examination
Survey, 1988-1994.
Diabetes Care.1998;21:518-524.Google Scholar 27.Godsland IF, Walton C, Felton C.
et al. Insulin resistance, secretion, and metabolism in users of oral contraceptives.
J Clin Endocrinol Metab.1992;74:64-70.Google Scholar 28.Chasan-Taber L, Willett WC, Stampfer MJ.
et al. A prospective study of oral contraceptives and non-insulin-dependent
diabetes mellitus among US women.
Diabetes Care.1997;20:330-335.Google Scholar 29.Singh KB. Menstrual disorders in college students.
Am J Obstet Gynecol.1981;140:299-302.Google Scholar