Is there an incremental rise in the risk of obstetric intervention with increasing maternal age?


Correspondence: Dr A. Rosenthal, Gynaecological Oncology Research Fellow, Gynaecology Cancer Research Unit, St. Bartholomew's Hospital, West Smithfield, London EclA 7BE, UK.


Objective To determine whether increasing maternal age increases the risk of operative delivery and to investigate whether such a trend is due to fetal or maternal factors.

Design Analysis of prospectively collected data on a maternity unit database.

Setting A postgraduate teaching hospital.

Population 6410 nulliparous women with singleton cephalic pregnancies delivering at term (37–42) weeks of gestation) between 1 January 92 and 31 December 95.

Main outcome measures Mode of delivery, rates of prelabour caesarean section, induction of labour and epidural usage.

Results There was a positive, highly significant association between increasing maternal age and obstetric intervention. Prelabour (P < 0.001) and emergency (P < 0.001) caesarean section, instrumental vaginal delivery (spontaneous labour P < 0.001; induced labour P= 0.001), induction of labour (P < 0.001) and epidural usage in spontaneous labour (P = 0.005) all increased with increasing age. In the second stage of labour fetal distress and failure to advance, requiring instrumental delivery, were both more likely with increasing maternal age (in both P < 0.001). Epidural usage in induced labour and the incidence of small for gestational age newborns did not increase with increasing maternal age (P = 0.68 and P = 0.50, respectively).

Conclusions This study demonstrates that increasing maternal age is associated with an incremental increase in obstetric intervention. Previous studies have demonstrated a significant effect in women older than 35 years of age, but these data show changes on a continuum from teenage years. This finding may reflect a progressive, age-related deterioration in myometrial function.


Many authors have reported increasing maternal age as a risk factor for poor obstetric outcome1; however, the majority of these studies have compared women over 35 or 40 years of age with a younger control group and few of have examined mode of delivery as a specific issue. Assuming that physiological changes associated with ageing are the cause of this poor outcome it would seem reasonable to assume that the effect would be a continual one and not merely confined to women over 35 years old. An extensive search of the literature failed to reveal any studies investigating a successive increase in poor outcome (in terms of mode of delivery) with increasing maternal age, as opposed to comparison of older and younger age groups.

The hospital at which this study was performed has a reputation for attracting professional women who may delay pregnancy until they have established their careers. Consequently, the mean age of a woman delivering at Queen Charlotte's is 31 years old. It would therefore seem that the population at this hospital is ideally suited to providing adequate numbers of women to investigate the hypothesis that increasing maternal age influences obstetric intervention.


The Queen Charlotte's maternity database has been in use since 1 January 1992 and contains detailed information on over 20,000 deliveries. Data are entered onto the database from specially designed partograms as follows. All normal deliveries are entered by midwives and all operative deliveries are entered by the senior house officer involved in the delivery. Each entry is then verified by the consultant in charge of labour ward (SPB) and discrepancies are clarified, and if appropriate, rectified.

The database from 1 January 1992 to 31 December 1995 inclusive was searched and 6410 anonymised, primiparous women were identified whose pregnancies were singletons, presenting cephalically, and who were delivered between 37 and 42 weeks. Data were then analysed for each year of maternal age. Women ≤ 16 or ≥ 40 years old were grouped together in ‘≤ 16’ and ‘≥ 40’ categories, as numbers at extremes of maternal age were very low. Small for gestational age newborns were classified using standard commercial growth tables for third centile values.

To test the hypothesis that outcome rates were associated with maternal age, logistic regression analysis was performed on the raw data. The significance of the association between maternal age and outcome rate was thus obtained. Odds ratios, together with 95% confidence intervals (representing the percentage increase in risk of a particular outcome for each increasing year of age), were also calculated. Goodness of fit statistics, reflecting how well the logistic regression modelled the assumption that outcome rates increased at a constant rate, year on year, were also calculated. All analyses were performed using STATA software version 5.


The mean maternal age at delivery was 29.4 years(range 11–48 years); the number of deliveries in each age group is shown in Table 1, along with the main outcome measures of mode of delivery, inductions, epidural usage and small for gestational age newborns, Figures 1 and 2 show rates of induction of labour and prelabour caesarean section, and caesarean section in spontaneous and induced labour, respectively. All increased with increasing maternal age (P < 0.001). The rates of instrumental vaginal delivery and the indications for them are shown in Figures 3 and 4, respectively. Again, all show highly significant trends with increasing maternal age: instrumental vaginal delivery in spontaneous (P < 0.001)and induced (P < 0.001) labour, and instrumental vaginal delivery for failure to advance in the second stage of labour (P < 0.001) and for fetal distress(P < 0.001).

Table 1.  Summary of outcome measures in each age group. Values are given as n (%). n = number of women with a particular outcome; %=percentage of total deliveries in stated age group with each outcome; PLCS = prelabour caesarean section; EMCS = emergency caesarean section in labour; SVD = spontaneous vaginal delivery; IVD = instrumental vaginal delivery; SGA = small for gestational age newborns.
Age (years)Deliveries nInductionPLCSSVDIVDEMCSEpiduralSGA
≤ 16378(22)0(0)29(78)7(19)1(3)28(76)3(8)
30654161(25)11(2)337 (52)231(35)75(11)470(72)41(6)
≥ 409034(38)11(13)26(29)38(42)15(17)62(67)14(16)
Figure 1.

Induction (▴) and prelabour caesarean (•) rates, plotted against maternal age.

Figure 2.

Emergency caesarean rates in induced (▴) and spontaneous (•) labour, plotted against maternal age.

Figure 3.

Instrumental vaginal delivery rates in induced (▴) and spontaneous (•) labour, plotted against maternal age.

Figure 4.

Instrumental vaginal delivery rates due to fetal distress (•) and failure to advance (▴), plotted against maternal age.

The use of epidural anaesthesia in all women at Queen Charlotte's and Chelsea Hospital is high (71%); age-related epidural usage is shown in Table 1 and Fig. 5. There was a significant increase in those women in spontaneous labour (P = 0.005) with increasing age, but the epidural rate in induced labour was unrelated to age. There was no association between the incidence of small for gestational age neonates and increasing maternal age (P = 0.50). Stillbirth numbers were too small for statistical analysis (n = 9: one stillbirth each for women at 24, 25 and 28 years of age, and two stillbirths each for women at 26,29 and 32 years of age).

Figure 5.

Epidural rates in induced (▴) and spontaneous (•) labour, plotted against maternal age.

The odds of a particular outcome increasing in frequency with each increasing year of maternal age are represented by the odds ratios (Table 2). For example, the odds ratio for instrumental vaginal delivery in spontaneous labour is 1.07, so the annual percentage increase in the rate of this outcome is 7% (ie, if the rate at 25 years was 27%, then the rate at 26 years would increase to 29%). The goodness of fit values indicate how well the logistic regression analysis models the assumption that the outcome rate increases year on year. Thus a goodness of fit P value of > 0.05 indicates that the assumption is likely to be real. Goodness of fit P values could not be calculated for some of the outcome measures for the following reasons: the numbers of women at each age were too small and/or the outcome rate was zero for some of the ages (eg, prelabour caesarean section) or the increase in outcome measure with increasing age was not linear (eg, instrumental vaginal delivery for failure to advance).

Table 2.  Odds ratios (95% CI) and goodness of fit (P) of an outcome measure increasing in frequency with each additional year of maternal age. PLCS = prelabour caesarean section; EMCS = emergency caesarean section in labour; IVD = instrumental vaginal delivery; FTA = failure to advance; FD = fetal distress; IOL = induction of labour.
OutcomeSampleOR (95% CI)P
  1. *A goodness of fit P value > 0.05 indicates that logistic regression adequately models the assumption that the outcome rate increases with increasing maternal age at a constant rate.

PLCSAll women1.16(1.08,1.19)
EMCSSpontaneous labours1.08(1.06,1.11)0.85*
EMCSInduced labours1.09(1.06,1.12)
IVDSpontaneous labours1.07 (1.02,1.07)
IVDInduced labours1.04(1.02,1.07)
IVD for FTAAll women1.05(1.03, 1.06)<0.001
IVD for FDAll women1.02(1.01,1.03)>0.2*
IOLAll women1.02(1.01,1.03)>0.2*
EpiduralSpontaneous labours1.03(1.02,1.05)>0.2*


It has long been known that mature nulliparous women are at increased risk of a variety of maternal and fetal complications including hypertension2, preeclampsia4, diabetes5, chromosomal abnormalities6 and stillbirth7. Given these facts, it is not surprising that fetal distress rates increase with maternal age8, as we found in our study (Fig. 4). It appears that fetal distress in our population was not a reflection of the proportion of newborns who were small for gestational age, as this latter outcome was unrelated to maternal age. This may reflect the fact that the ‘fetal distress’ category entered onto the database is the subjective opinion of the junior doctor who performed the delivery. This is supported by the observation that failure to progress in the second stage (arguably a more objective endpoint), as a cause of instrumental vaginal delivery, increased proportionally much more with increasing maternal age than did fetal distress (Fig. 4). There were too few stillbirths in our population to carry out any statistical analysis, however, it is interesting to note that none occurred in women over the age of 32 years.

There is a wealth of information indicating increased caesarean section rates in older women, but not so much evidence supporting increased instrumental delivery rates1. It seems a reasonable hypothesis that the uterus, like all organs undergoes a process of ageing, and as a spontaneous vertex delivery is largely dependent on adequate uterine contraction, one might expect the mechanical function of the uterus to deteriorate with increasing age, resulting in failure to advance in the second stage of labour and consequently increased intervention rates. Certainly, the incidence of postpartum haemorrhage9 and the length of the second stage of labour10 (irrespective of epidural usage) have been found to increase gradually with increasing maternal age in nulliparae, and these observations would support the concept of myometrial deterioration.

There has been the suggestion that women aged over 35 have increased oxytocin requirements11 which would also support the concept of a continuum of myometrial deterioration. Ezra et al.2 demonstrated increasing intervention rates in five year age groups from 15–40 years; although statistics were quoted only for the comparisons of women younger than 35 and older than 35. The authors felt that intervention rates were driven up by obstetricians' and maternal anxiety, despite the fact that no analysis of indication for intervention was performed and intervention rates appeared to increase with each age group from 15 to 20 years upwards, rather than just for those women older than 35 years. The authors' explanation for increased intervention being due to obstetrician and patient anxiety was the relatively smaller increase in perinatal morbidity and the absence of significantly increased perinatal mortality with increasing maternal age which their study showed. Although there is evidence that anxiety in primiparae and their obstetricians increases with increasing maternal age12, we would dispute that this is the prime cause of increased intervention rates in this group of women.

To the best of our knowledge, our study is the first to demonstrate a continuing increased risk of operative delivery with increasing maternal age, suggesting a decline in uterine function. Although epidural usage has been found to be associated with increased operative delivery10 which helps explain the high overall intervention rates in our study, the group of women having induced labour, in whom epidural usage did not increase with age, still had an increased risk of intervention, demonstrating an independent effect of age. Although the nature of the statistical test used is such that a sudden change in the rate of an outcome from one age to another could not be predicted well (eg, the high rates of instrumental vaginal delivery in women < 17 years labouring spontaneously), it is clear from Figs 1–5 that the general association between maternal age and outcome is real and, for some outcomes, well-modelled by the logistic regression analysis (which assumes that outcome rates increase at a constant rate). This is further supported by the goodness of fit statistics (Table 2), where P is > 0.05 for four out of the nine outcomes significantly associated with maternal age. Where the goodness of fit was > 0.05 (eg, induction rates), it can be concluded that the logistic regression models the data adequately and the outcome rate does increase at a constant rate with each increasing year of maternal age. The outcomes where the goodness of fit was < 0.05 or unobtainable were still associated with maternal age; however, the relation was either nonlinear (eg, instrumental vaginal delivery for failure to advance) or numbers and outcome rates were too low to enable calculation (eg, prelabour caesarean section). Some of these outcomes are better modelled by nonlinear equations eg, the graph for instrumental vaginal delivery for failure to advance (Fig. 4) appears to be linear from age 22 to age 32, but outside this range, the linear model would overestimate outcome rates. This may be due to lower numbers of women at the extremes of age, the grouping together of women ≤ 16 and ≥ 40, or a genuine biological effect. Nevertheless, it is clear from the general trends seen in Figs 1–5 and the significance of the relation between the outcome measures and maternal age that obstetric intervention is influenced in a continuous fashion by maternal age. Moreover, for the outcomes where goodness of fit indicates adequate logistic regression modelling, it is possible to predict, with reasonable accuracy, the rate of an outcome at a particular age.

We have come across two studies which provide histopathological evidence to support the hypothesis of deteriorating myometrial function with increasing age. Popov et al.13 observed that with advancing age in primiparous women, there is progressive thickening of the muscular layer of myometrial arteries due to fibrosis. More recently, Crawford et al.14 examined uterine arteries in 59 premenopausal (mean age 41.6, SD ±6.6 years) and 15 postmenopausal women (mean age 66.8, SD ± 9.9 years) not taking hormone replacement therapy (parity was not described). In the premenopausal group, 55.9% had intimal thickening, 40.7% had simple plaques and 3.7% had complex plaques. There was a significant linear correlation between increasing age and the intimal area: medial area ratio in both the premenopausal (P = 0.04) and postmenopausal women (P = 0.01), indicating age-related progressive thickening of the intima. It would therefore appear that the uterus is subject to pathological processes associated with ageing and presumably, these might reduce its ability to contract.

It is now well recognised that in the developed world, women are choosing to delay child bearing until they are well into their third decade3,15 reflecting the advent of reliable contraception and greater equality in the workplace. This trend shows no signs of slowing and has major implications. Not only are subfertility rates likely to increase, but so are perinatal and maternal morbidity rates associated with more mature nulliparae. This group of women are frequently well informed and have a great deal of knowledge about labour and its complications. Although the absolute risk of intervention increases only fractionally with each increasing year of maternal age and predicting an individual's risk of an outcome on the basis of age alone is not possible, it would still appear reasonable to inform women about the general increased risk of operative delivery associated with advanced maternal age.

In addition to patient information, there is clearly also a resource allocation issue here. Obstetric units in parts of the country where women delay pregnancy might quite reasonably expect higher costs in terms of looking after increased numbers of women having operative vaginal delivery and caesarean sections. Furthermore, the justification of operative delivery rates is a pressure under which all units are working. The age of the population concerned is influential and should be taken into the equation.

Despite the extremely high rate (12%) of prelabour caesarean section among women who are 40 years and older, which probably reflects a degree of anxiety in obstetricians and the mothers, we believe that the incremental rise in risk of intervention observed in our women demonstrates a genuine reduction in performance in labour.


The authors would like to thank Ms A. Gosling (Department of Medical Statistics and Evaluation, The Royal Postgraduate Medical School, London) and MS J. Thomas (Department of Medical Statistics, St Bartholomew's-Hospital, London) for their help with the statistical analysis.