GCS Smith, MD, PhD, Professor of Obstetrics and Gynaecology, Cambridge University, Box 223, Rosie Maternity Hospital, Robinson Way, Cambridge, CB2 0SW, UK. Email firstname.lastname@example.org
Objective The risk of operative delivery at term increases linearly with age at first birth. It has been hypothesised that this is because of a deleterious effect of a prolonged interval between menarche and first birth on uterine function. The aim of this study was to test a prediction from the hypothesis, namely, that the risk of operative first delivery would decline with later age at menarche.
Design Retrospective analysis of a prospective cohort study.
Setting The ALSPAC prospective cohort study enrolled pregnant women resident in Avon, UK with expected dates of delivery from 1 April 1991 to 31 December 1992.
Population A total of 3739 primipara recruited to the ALSPAC cohort who experienced labour at term with a singleton infant in a cephalic presentation.
Main outcome measure Operative delivery, defined as caesarean section or operative vaginal birth.
Result The rate of operative delivery was highest among women with age at menarche in the bottom quartile (32.4%, menarche aged ≤12) and was lower in the second (30.3%, menarche aged 13), third (29.2%, menarche aged 14) and top (26.9%, menarche aged ≥15) quartiles (test for trend, P = 0.01). When adjusted for height, body mass index, marital status, smoking status, induction of labour, week of gestation of delivery and birthweight percentile; the odds ratio for operative delivery associated with a 5-year increase in age at menarche (0.78, 95% CI 0.61–0.99) was very similar to the odds ratio for a 5-year decrease in age at delivery (0.73, 95% CI 0.67–0.79). There was no association between age at menarche and the risk of operative delivery following adjustment for the interval between menarche and the first birth (adjusted odds ratio 0.98, 95% CI 0.77–1.25).
Conclusion Later menarche is associated with a decreased risk of operative delivery by decreasing the interval between menarche and first birth. The observation is consistent with the hypothesis that prolonged hormonal stimulation of the uterus prior to the first birth has a deleterious effect on uterine function.
Many studies have demonstrated that the risk of caesarean delivery was increased among older parturients. Some have suggested that this observation may reflect physician or maternal preference1,2 and others that it may reflect a true biological effect of aging.3,4 The risk of emergency caesarean delivery, the risk of operative vaginal delivery and the duration of labour all increase linearly with age, from 16 years upwards.3–5 It has been suggested that the presence of a linear relationship across the whole range of age at first birth cannot be explained by physician or maternal preference and is most likely to represent a true biological effect of aging.3 Consistent with this, it has been shown that the spontaneous contractility of isolated human myometrial strips in vitro declines with advancing maternal age.5 Collectively, these observations are consistent with a biological effect of aging on uterine function. However, it is currently unclear why advancing age across the range of the reproductive years might impair uterine function.
Women who delay childbirth will experience prolonged exposure to the female sex hormones prior to their first birth. Depending on the contraceptive method employed, these could be endogenous (for example, women using barrier methods, non-hormonal intrauterine devices and sexual abstinence) or exogenous synthetic derivatives of progesterone and estrogen (for example, women using the combined pill). Uterine smooth muscle (myometrium) and the cervix express estrogen and progesterone receptors6,7 and many aspects of uterine function are controlled by the levels of these hormones (see review8). It had previously been hypothesised that the association between increasing age at first birth and dysfunctional labour may be related to an adverse effect of prolonged pre-pregnancy stimulation of the uterus by estrogen and progesterone.5 The aim of the present study was to test a prediction arising from the hypothesis, namely, that later age at menarche would be independently predictive of a lower risk of operative first delivery.
The ALSPAC prospective cohort study enrolled pregnant women resident in Avon, UK with expected dates of delivery from 1 April 1991 to 31 December 1992. The study design is described in detail elsewhere.9 Ethical approval for the study was obtained from the ALSPAC Law and Ethics Committee and the Local Research Ethics Committees. Women recruited to the study completed questionnaires in the prenatal period. Pregnancy outcome data were obtained for cohort members delivering at Southmead or St Michael’s Hospitals by linking the ALSPAC data to the hospitals’ computerised birth records (both hospitals employed the same maternity database, ‘STORK’). These data were not available for women who delivered in the other local hospital (Weston General) or who delivered at home.
Ethnicity, pre-pregnancy weight, height, marital status, parity and smoking status were all obtained from prenatal questionnaires. Gestational age was defined using the last menstrual period with adjustment for ultrasound estimation of gestation where this was available. The birthweight was taken from the infant’s birth record and was converted into a percentile for gestational age using previously described data.10 Cephalic presentation was defined as any case where the presentation was by the occiput, brow or face and where both the mode of delivery and indication for operative delivery were consistent with cephalic presentation. Labour was defined by a non-missing value for duration of labour and where the mode of delivery was not elective caesarean delivery. Operative vaginal delivery was defined as delivery using forceps or vacuum (ventouse) and all operative delivery was defined as operative vaginal or caesarean delivery. The computerised birth record was compared with the clinical notes in a random sample of 50 women from each hospital in both 1991 and 1992. This demonstrated overall rates of error in coding of method of labour onset in 6%, presentation in 2% and mode of delivery in 3%. Checks of internal consistency of the hospital database also indicated low rates of error. Among 11 089 records documenting cephalic vaginal delivery, only 23 (0.2%) had a missing value for duration of labour. Among 542 records documenting delivery by elective caesarean, there was only a single record (0.2%) with a documented duration of labour. Among 11 883 records documenting cephalic presentation, only 27 (0.2%) documented a mode of delivery that was indicative of non-cephalic presentation.
The inclusion criteria for the current study were nulliparous women, delivering a singleton, liveborn infant in a cephalic presentation, in labour at term, who had a documented age at menarche. Records with any of the above internal inconsistencies were excluded.
Continuous variables were summarised by the median and inter-quartile range and groups were compared using the Kruskal–Wallis test. Univariate comparisons of categorical data were performed using the χ2 test for trend. All P values were two-sided, and a P value <0.05 was considered significant. Adjusted odds ratios were estimated using multivariate logistic regression.11 Analysis of the association between age at menarche and the risk of operative birth was performed by categorizing age at menarche into quartiles. In further analyses, to compare the relationship between age at menarche and age at delivery, both variables were treated as continuous. When treated continuously, age-related variables were divided by 5 to result in a more precise estimate of the association when odds ratios were expressed to 2 decimal places. Age at delivery was expressed as the odds ratio for a 5-year decrease in age to allow direct comparison with age at menarche. Approximately 0.5% of values for age at menarche were outside the range 9–17: values below and above this range were truncated as 9 or 17 respectively. Approximately 0.7% of values for age at delivery were outside the range 16–40: values below and above this range were truncated as 16 or 40 respectively. Linearity in the log odds scale was assessed using fractional polynomials.12 Interactions were tested using the likelihood ratio test. The principal method for treatment of missing data was multiple imputation using chained equations.13 To assess whether the method of treatment of missing variables influenced the analysis, the results of multivariate analysis using this method was also compared with using dummy variables to indicate missing data and replacing missing values with the median (continuous) or most common (categorical) value for the population. The inter-relationships between age at menarche, duration of labour, the interval between menarche and first birth, body mass index and height were analyzed using multiple linear regression. All of the above statistical analyses were performed using the Stata software package (Stata Corporation, College Station, TX, USA), version 10.1. The power of the study to assess the relationship between age at menarche and mode of delivery was assessed using the SD of the age at menarche within the study cohort. Power was calculated using the assumption that a given magnitude of increase in age at menarche would have the same proportional effect on the odds of operative delivery as the same decrease in age at the time of the first birth. Power calculations were performed using nQuery Advisor version 7.0 (Statistical Solutions, Saugus, MA, USA).
The ALSPAC study recruited 14 541 pregnant women who either returned at least one questionnaire or attended a ‘Children in Focus’ clinic by 19/7/99. There were 9997 (68.2%) records of a liveborn, singleton pregnancy, with a mode of delivery recorded on the STORK database and with a documented age at menarche. Among this group, 4382 (43.8%) were documented as nulliparous. Among these women, 284 (6.5%) delivered outside the range 37–43 weeks, 369 (8.4%) did not document the infant as being in a cephalic presentation and 216 (4.9%) had no documented duration of labour. A total of 643 (14.7%) records had one or more of these exclusions leaving 3739 eligible women who were the study group for the present analysis.
The distribution of age at menarche and age at first birth are illustrated (Figure 1). The mean age at menarche was 12.9 and the SD was 1.55. Age at menarche was positively associated with height and negatively associated with body mass index (Table 1). There was a linear decline in the risk of operative delivery with increasing quartile of age at menarche (Figure 2). The rate of operative delivery was highest among women with age at menarche in the bottom quartile (32.4%, menarche aged ≤12) and was lower in the second (30.3%, menarche aged 13), third (29.2%, menarche aged 14) and top (26.9%, menarche aged ≥15) quartiles (odds ratio for 1 quartile increase 0.92, 95% CI 0.86 to 0.98, test for trend, P = 0.01). Adjustment for other maternal characteristics, with missing data handled using multiple imputation, had a minimal effect on the association between age at menarche and the risk of operative delivery (adjusted odds ratio for 1 quartile increase 0.93, 95% CI 0.87 to 1.00, test for trend, P = 0.04, Table 2). When the composite outcome was divided into its two components, a one quartile increase in age at menarche was associated with an odds ratio for caesarean section of 0.94 (95% CI 0.82–1.08, P = 0.39) and an odds ratio for assisted vaginal delivery of 0.92 (95% CI 0.86–0.99, P = 0.02).
Table 1. Characteristics of the cohort by quartile of age at menarche
Maternal characteristics and outcome
Quartile of age at the time of menarche*
1 (n = 1503)
2 (n = 1102)
3 (n = 644)
4 (n = 490)
IQR, inter-quartile range.
Data were missing from (n) records in relation to height (61), body mass index (254), marital status (86), ethnicity (138) and birthweight percentile (47).
*Numbers in quartiles are unequal due to ties.
**Chi-square test for trend or Kruskal–Wallis test, as appropriate. Data expressed as n (percent) unless stated otherwise.
Age at menarche (range)
Age at delivery
Body mass index
Week of gestation
Onset of labour
Duration of labour
Mode of delivery
Table 2. Unadjusted and adjusted odds ratios for operative delivery in relation to maternal characteristics
Odds ratio (95% CI)
Odds ratio (95% CI)
CI, confidence interval.
*Adjusted for quartile of age at menarche (categorical), maternal age, height, body mass index, smoking status, ethnicity, marital status, induced labour, gestational age at birth and birthweight percentile. Missing data handled by multiple imputation using chained equations.
Quartile of age at menarche
Age at delivery
Body mass index
Onset of labour
Week of gestation
A series of analyses was then conducted where age at menarche and age at delivery were treated as continuous variables. Logistic regression analysis using multiple fractional polynomials demonstrated that the association between operative delivery and both age at menarche and age at delivery were linear in the log odds scale. The protective effect of a 5 year increase in age at menarche was very similar to the protective effect of a 5 year decrease in age at the time of the first delivery (Table 3). Adjustment for other maternal characteristics has a minimal effect on the nature and statistical significance of the associations. Three methods for handing missing data yielded virtually identical results. The association between operative delivery and a 5-year increase in age at menarche was unchanged if cases with extreme values of menarche were excluded rather than truncated (unadjusted odds ratio 0.72, 95% CI 0.60–0.96, P = 0.02). There was no significant interaction between age at menarche and whether the onset of labour was spontaneous or induced (odds ratio for interaction term 0.86, 95% CI 0.51–1.48, P = 0.59).
Table 3. Analysis of age at menarche and age at birth as continuous variables
OR for 5 year increase in age at menarche (95% CI)
OR for 5 year decrease in age at delivery (95% CI)
CI, confidence interval; OR, odds ratio.
Adjusted odds ratios using three methods for treating missing data: 1 using multiple imputation using chained equations, 2 by assigning median or most common value to missing value, 3 by creating dummy variables for missing value and including in the model.
*Adjusted for height, body mass index, smoking status, ethnicity, marital status, induced labour, gestational age at birth and birthweight percentile.
The median interval from menarche to first delivery was 14 years (inter-quartile range 11 to 17). This interval was inversely correlated with age at menarche (r2 = 0.06, P < 0.001). Adjustment for the duration of this interval resulted in complete loss of the association between age at menarche and the risk of operative delivery, but had no effect on the associations between age at menarche and the pre-pregnancy body mass index or height (Table 4).
Table 4. The effect of adjustment for the duration of the interval between menarche and first delivery on the associations with age at menarche
Associations with a 5-year increase in age at menarche
Unadjusted (95% CI)
Adjusted* (95% CI)
Body mass index is expressed in kg/m2, height is expressed in cm. Odds ratios estimated using logistic regression and coefficients by linear regression.
*Adjusted for the interval between menarche and first birth only.
Body mass index
−1.69 (−2.07 to −1.31)
−1.58 (−1.97 to −1.19)
2.40 (1.69 to 3.11)
The duration of labour did not differ across the quartiles of age at menarche. The relationship between age at menarche, age at delivery and duration of labour was explored using linear regression. A 5-year increase in age at menarche was associated with −0.09 hour difference in the duration of labour (95% CI −0.61 to 0.43, P = 0.7). A 5-year decrease in age at delivery was associated with a −0.33 difference in duration of labour (95% CI −0.49 to −0.16, P < 0.001).
Power calculations were performed with the assumption that a 1 SD increase in age at menarche would be associated with an odds ratio of 0.879, based on a previous analysis.5 Setting alpha at 0.05 (two-sided), the size of the study group yielded 38% power for an outcome with 5% frequency, 89% power for an outcome with 25% frequency and 92% power for an outcome with 30% frequency. The frequency of caesarean delivery in the study group was 5.4%, the frequency of operative vaginal delivery was 25.2% and the frequency of the composite outcome of all operative delivery was 30.5%.
The main finding of the present study is that the risk of operative delivery at term among nulliparous women decreased with later age at menarche. The association was linear and persisted after adjusting for potential confounders, such as maternal height and body mass index. The relationship between age at menarche and the risk of operative delivery was because of the fact that age at menarche in part determined the duration of the interval from menarche to the first birth (also known as the gynaecologic age14). Adjustment for this interval resulted in complete loss of the association between age at menarche and the risk of operative delivery. Hence, it is concluded that later menarche is associated with a decreased risk of operative delivery by decreasing the interval between menarche and first birth.
It had previously been hypothesised that increasing age at the time of first birth increased the risk of operative delivery by increasing the interval between menarche and first birth.5 This hypothesis predicts that a woman’s age at menarche would be negatively associated with her risk of operative delivery. Three features of the present analysis support this hypothesis. First, the relationship between age at menarche and the risk of operative delivery was independent of other maternal characteristics. Second, the proportional effect of a given increase in the age at menarche was similar to a given decrease in age at first birth. Third, the association between age at menarche and the risk of operative delivery was no longer apparent when adjusted for the interval between menarche and first birth. The current analysis provides strong evidence in support of the hypothesis. Hence, this analysis is significant in that it is the first description of the association between age at menarche and the risk of operative birth. However, the key significance of the current analysis is that it suggests a mechanism linking delayed childbirth to the risk of operative delivery, that is that the true mediator of the associations between the risk of operative delivery and both age at menarche and age at first birth is the duration of the interval between menarche and first birth. The public health importance of this is that delaying childbirth is a major determinant of recent rises in the caesarean section rate.5
A prolonged interval between menarche and first birth will result in prolonged pre-pregnancy stimulation of the uterus by estrogen and progesterone, with the exact pattern being determined by the contraceptive method employed. The most common cause for operative delivery among women having their first birth during labour at term is poor progress during labour.15,16 The current analysis is consistent with the hypothesis that prolonged pre-pregnancy stimulation of the uterus by estrogen and progesterone adversely affects the uterus in a way that impairs its function during labour. The hypothesis has parallels in relation to other disorders of the female reproductive organs, in particular the risk of breast cancer. The risk of this condition is also decreased with later age at menarche. This association is thought to be because of an adverse effect of prolonged stimulation by mammotrophic hormones, principally estrogens and progestogens.17 Given that the uterus is also profoundly under the control of these hormones,8 it is biologically plausible that prolonged pre-pregnancy stimulation of the uterus by estrogen and progesterone could also adversely affect uterine function.
The prospective design of this study means that women were asked about their age at menarche prior to delivery, precluding the possibility of recall bias in relation to ultimate mode of delivery. A further strength of this study was the availability of detailed information on relevant potential confounders. Early menarche was associated with short stature and obesity and both are recognised to be risk factors for operative delivery.18 Although the association was largely unchanged by multivariate analysis, the persisting association may be explained by residual confounding. In particular, body mass index is a proxy measure of adiposity and it is possible that residual confounding by obesity may explain the results observed. This question was addressed by examining the effect of adjusting associations between age at menarche for the interval between menarche and first birth. It was observed that adjustment for this interval abolished the association between age at menarche and the risk of operative delivery (Table 4). By contrast, adjustment for the interval between menarche and first birth had no material effect on the association between age at menarche and either maternal height or BMI. This analysis demonstrates that age at menarche is associated with the risk of operative delivery by its contribution to the interval between menarche and first birth. Moreover, it indicates that the association between age at menarche and maternal anthropometric characteristics is mediated by a different mechanism.
A relative weakness of this study is that these data were collected in 1991–1992. Consequently, the rates of operative delivery were relatively low, with a caesarean delivery rate of 5.4%. Power calculations demonstrated that the study had approximately 90% power to detect an effect of age at menarche on the risk of all operative delivery, but only approximately 40% power to detect an association with caesarean delivery. Hence, the study was underpowered to detect an effect of age at menarche on caesarean section risk alone but was adequately powered to detect an effect on the composite outcome of operative delivery. It may be argued that the composite outcome is determined by many factors, such as fetal distress and maternal exhaustion. However, the primary aim of the present study was to shed light on the relationship between maternal age and the risk of operative delivery. In a previous study of nulliparous women in labour at term with an infant in a cephalic presentation, we demonstrated that a 5-year increase in maternal age was associated with an odds ratio for assisted vaginal delivery of 1.49 (95% CI 1.48–1.50) and an odds ratio for emergency caesarean section of 1.49 (95% CI 1.48–1.51).5 Given the virtually identical relationships between maternal age and the risk of both assisted vaginal delivery and emergency caesarean section, we conclude that the composite outcome is appropriate given our primary interest in the effects of maternal age.
The STORK database lacked validated information on the indication for delivery. However, a prospective analysis of the indication for primary intrapartum caesarean section in nulliparous women at term in 1991 demonstrated that over half of these procedures with the infant in a cephalic presentation were due to failure to progress.19 Although we did not observe significant variation in the duration of labour in association with age at menarche, linear regression analysis demonstrated wide confidence intervals and these included the point estimate for the association with maternal age. Duration of labour is difficult to define and any retrospective analysis of data collected for other purposes is likely to face problems with the reliability of indicators of progress in labour. We are addressing this by conducting a prospective cohort study of unselected nulliparous women, which addresses some of the weaknesses of the present analysis, in that we are collecting detailed information on contraceptive history, in addition to age at menarche, and outcome data is being collected in a systematic fashion and includes information on duration and dose of oxytocin use.20
The public health importance of this finding is that population trends in delaying childbirth appear to be an important determinant of recent increases in caesarean delivery rates.5 The present analysis suggests a likely mechanism for this association. Moreover, age at menarche has fallen over recent years.21 The current data suggest that a consequence of this change may be to increase the incidence of poor progress during labour and, hence, population trends of earlier menarche may also have contributed to recent rises in rates of caesarean delivery. Mechanistic understanding of the basis for this effect could possibly identify whether different methods of hormonal contraception might have different effects on the deterioration in uterine function among women who wish to delay childbirth. For example, it is possible that any adverse effect on the uterus of prolonged pre-pregnancy use of exogenous synthetic estrogen and progesterone for contraception may differ in relation to cyclical or noncyclical administration.
Disclosure of interests
The author has no competing interests of any form in relation to this work.
Contribution to authorship
The sole author is wholly responsible for all aspects of this study.
Details of ethics approval
The analysis is covered by the ALSPAC ethical approval, given by Bristol and Weston Health Authority (reference E1808) on 28 November 1989.
The data extraction was funded through NIHR Senior Investigator funding to GCSS and financial support for the analysis (hardware and software) was provided through the NIHR Cambridge Comprehensive Biomedical Research Centre.
I am extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them and the whole ALSPAC team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists and nurses. The UK Medical Research Council, the Wellcome Trust and the University of Bristol provided core support for ALSPAC. I am grateful to Scott Nelson (Professor of Obstetrics & Gynaecology, Glasgow University) for suggesting use of the ALSPAC cohort and to Dr Angela Wood (Lecturer in Biostatistics, Department of Public Health, Cambridge University) for discussions on the statistical analysis.
1. Objective and background:
Is the study hypothesis well stated and did the author provide a convincing argument for the proposed study?
2. Study design:
What is the design of the study? What constitutes a retrospective study design? Is the study population which constitutes the ALSPAC cohort similar to yours? Are the outcome measures well defined and is there any risk of misclassification of the primary exposure (age at menarche) or operative delivery? Was the effect of hormonal contraceptive methods well explored by the study? Can the use of oxytocin or epidural bias the results, if so, in what way? Were appropriate statistical methods used in the data analyses?
3. Clinical implication:
Based on the findings of the study, how will you counsel women on the impact of age at menarche on the risk of operative delivery? Can the findings from this study be correlated with the impact of delaying childbirth on the risk of operative delivery? Overall, is this a hypothesis generating or a practice changing study?
4. Future research:
If you want to reproduce the study findings, what changes in design would you make?
Division of Maternal Fetal Medicine, Department of Obstetrics and Gynecology, 4990 Children’s Place, St Louis, MO, USA Email email@example.com