Dr C Greenwood, Department of Obstetrics and Gynaecology, Level 4, The Women's Centre, John Radcliffe Hospital, Oxford OX3 9DU, UK. Email email@example.com
Objective To evaluate the effect of a first trimester ultrasound dating scan on the rate of induction of labour for prolonged pregnancy.
Design Randomised controlled trial to include 400 women in each arm of the trial.
Setting Participating general practices and a district general teaching hospital.
Population Women attending their general practitioner in the first trimester to confirm pregnancy, in whom a first trimester ultrasound scan was not indicated.
Methods Women randomised to the study group (scan group) underwent an ultrasound dating scan between 8 and 12 weeks, measuring crown-rump length. The estimated date of delivery (EDD) was changed if there was a discrepancy of more than 5 days from the gestation, calculated from the last menstrual period (LMP). For the remaining women (no-scan group), gestation was determined using the LMP.
Main outcome measures The rate of induction of labour for prolonged pregnancy.
Results Due to circumstances beyond the researchers’ control, recruitment was abandoned when 463 women had been enrolled. The EDD was adjusted in 13 (5.7%) women in the scan group and in 2 (0.9%) in the no-scan group. There was no difference in the rate of induction for prolonged pregnancy between the scan (19 [8.2%]) and the no-scan (17 [7.4%]) groups (relative risk 1.10; 95% CI 0.59–2.07).
Conclusions Acknowledging the reduced numbers recruited for study, it is concluded that there is no evidence that a first trimester ultrasound dating scan reduces the rate of induction of labour for prolonged pregnancy and may result in a more expensive healthcare strategy.
Prolonged pregnancy is the largest single indication for induction of labour in the UK.1 The accurate assessment of gestational age is essential if unnecessary interventions based on inaccurate estimate of gestation are to be prevented.
Induction of labour after 41 weeks of gestation is routine in most obstetric units in the UK; in Oxford, up to 10% of labour inductions are for this indication. According to one study, induction of labour is associated with a longer inpatient stay, a shorter but more intensively monitored time on the delivery ward and a higher intervention rate.2
Estimation of gestational age based on the last menstrual period (LMP), even in those pregnancies where menstrual history is certain, has been shown to be unreliable.3–5 Compared with the use of certain menstrual dates, ultrasonographic dating has been shown to lead to up to a 70% reduction in the number of pregnancies considered postterm.6,7 The Cochrane Systematic Review of randomised controlled trials of ultrasound for fetal assessment in early pregnancy8 and of interventions for preventing or improving the outcome of delivery at or beyond term9 shows that accurate calculation of gestational age by early ultrasound and subsequent adjustment of delivery date reduces the incidence of women requiring induction of labour for apparently postterm or prolonged pregnancy. However, we are aware of only two previous randomised controlled trials addressing this issue, where a dating scan was performed at less than 15 weeks of gestation.10,11 In the first of these, Ewigman et al.10 reported a large trial of 915 women randomly assigned to receive either a single routine screening ultrasound or an usual prenatal care. They showed no differences between the groups in inductions for postdates pregnancy (about 7% in both groups), despite the estimated date of confinement being altered in almost a quarter of routine ultrasound patients. In contrast, Bennett et al.11 reported a smaller trial of 196 women randomly assigned to either an additional first trimester scan or a second trimester scan only. A high proportion of the intervention group (41.3%) had their gestational age adjusted on the basis of the crown rump length (CRL) measurement, and following this, there was a significant reduction in the rate of labour induction for postterm pregnancy (12/92 versus 5/104; P= 0.04).
Ultrasound measurement of CRL in the first trimester of pregnancy has been shown to be the most accurate method of estimating gestation.12 The National Institute for Clinical Excellence (NICE) guidelines for routine antenatal care for the healthy pregnant women recommend that women should be offered an ultrasound examination ideally between 10 and 13 weeks using a CRL measurement to determine gestational age.13 The estimation of fetal size to determine gestational age has been shown to improve the performance of midtrimester serum screening for Down syndrome.14,15
The purpose of this study was to determine if the calculation of gestational age by a first trimester ultrasound scan decreased the number of women undergoing induction of labour for prolonged pregnancy, and if so, this had attendant cost savings.
Women who consented to participate were recruited to the trial between February 1999 and October 2001, at their first appointment with their general practitioner, to confirm the pregnancy, at one of 20 collaborating general practices. They were eligible if they were in the first trimester of pregnancy and had no obstetric indication for a first trimester ultrasound examination. Demographic details were recorded at recruitment, and outcome data were collected following discharge from hospital, after confinement. Approval for the study was obtained from the Central Oxford Research Ethics Committee.
Participating general practices were provided with a series of consecutively numbered, sealed, opaque envelopes randomised in blocks of six, which allocated the women to the ‘scan’ or ‘no-scan’ group. All women were offered an anomaly ultrasound examination at 20 weeks of gestation.
Women allocated to the ultrasound group had their details registered by telephone with the study coordinator, and an appointment was arranged for an ultrasound examination to measure the CRL between 8 and 12 weeks of gestation. The ultrasound examination was usually performed abdominally, but where indicated, a vaginal scan was performed in addition, using a Toshiba SSH 140A machine, equipped with 5 MHz curvilinear and 6 MHz transvaginal transducers. All the scans were performed by one of two staff, either by a senior sonographer in the prenatal diagnosis department or by a fetomaternal subspecialist obstetrician. The CRL measurement was repeated until three values within 1 mm of each other were obtained, and the greatest was used as the CRL. The estimated date of delivery (EDD) was calculated using Naegele's rule (280 days from the start of the LMP, dependent on cycle length).16 For women randomised to scan, the EDD was recalculated if the scan dates differed by more than 5 days with the menstrual dates. Previous research suggested 95% reliable prediction within 5 days.12 The EDD was entered into the patient's obstetric notes, and all subsequent management decisions were based on this assessment of gestational age. Multiple gestations were disclosed as were suspected structural fetal anomalies with referral to the prenatal diagnosis unit. Nonviable or ectopic pregnancies were managed appropriately.
Power calculations were based on a reduction in induction of labour for prolonged pregnancy in Oxford from 10 to 5% with the use of first trimester dating scans. Based on this, an estimated 400 women were required in each arm of the trial with significance (2α) = 0.05 and power (1-β) = 0.8. Private first trimester nuchal scanning for Down syndrome screening was introduced during the recruitment phase of this trial, which compromised the rate of enrolment. Despite doubling the time allocated for the trial, it was not possible to enlist the required number of women, and eventually, the trial was stopped when 463 women had been recruited because of staffing constraints. Because of the difficulties with recruitment, when the outcomes on 200 women were known, an unscheduled interim analysis was performed. Data were analysed by a third party, and following this, advice was given to continue to 500. The findings were not revealed to those conducting the trial.
Analysis of the data from that time on using the primary outcome measure (induction for prolonged pregnancy) to investigate optimal sample size17 shows 6/104 (6%) with induction for prolonged pregnancy in the intervention group and 8/96 (8%) in the control group. Using these proportions and the same criteria for resetting sample size (18) (significance [2α] = 0.05 and power [1-β] = 0.8), the equivalent target sample size would have been 5306 cases. Using these same proportions and the same significance level (2α) = 0.05, the power of the study, if 800 were recruited (as originally planned), was only (1-β) = 0.16. It was therefore unlikely that the intervention would show a difference in postterm induction that could be considered either statistically significant or clinically worthwhile whether 800 or 500 were enrolled.
At the same stage, using spontaneous onset of labour as an outcome, the data showed 70/104 (68%) with spontaneous labour in the intervention group and 53/96 (55%) in the control group. Using the same criteria for setting sample size again, the total equivalent sample size for this outcome measure was 468 cases.
The data were analysed according to intention to treat. Statistical analyses were performed with SPSS for windows statistical package (version 11; SPSS, Chicago, IL, USA). Results in the two groups were compared by chi-square test statistic and Student's t test.
A total of 463 women were recruited between February 1999 and October 2001. Two hundred and thirty-three women were randomised to the scan group and 230 to the no-scan group; ultimately, there were 202 and 191 women available for analysis in the respective groups (1Figure 1). As shown in (1Table 1), the two groups were comparable with respect to demographic characteristics. The EDD was adjusted in 13/233 (5.6%) women in the scan group and in 2/230 (0.9%) women in the no-scan group (relative risk [RR] 1.77; CI 1.42–2.21). There was one intrauterine death in the first trimester scan group at 24 weeks of gestation due to fetal triploidy. This pregnancy was redated at the dating scan, although this size discrepancy probably reflected early intrauterine growth restriction. There were no intrauterine deaths in the no-scan group.
Table 1. Demographic details
Scan group, n= 233
No-scan group, n= 230
Values are expressed as n (%) and mean (SD).
Gestational age at booking (days)
Midwifery care to shared care
159 (78):45 (22)
There was no difference in the rate of induction for prolonged pregnancy between the scan (19/233 [8.2%]) and the no-scan (17/230 [7.4%]) groups (RR 1.10; CI 0.59–2.07). The proportion of women who laboured spontaneously was greater in the scan group (154/233[66.5%]) than in the no-scan group (132/230 [57.4%]). The overall rate of induction of labour was greater in the no-scan group (34 [14.6%]) versus in the scan group (46 [20.0%]). Neither of these findings reached statistical significance. Also there were no significant differences in the number of assisted deliveries or caesarean sections between the two groups (2Table 2).
Table 2. Onset of labour and mode of delivery
Scan group, n= 233
No-scan group, n= 230
RR (95% CI)
Total number of elective deliveries
Labour induction for prolonged pregnancy
Total labour inductions
Caesarean section before labour
Spontaneous onset of labour
Spontaneous vaginal deliveries
Assisted vaginal deliveries
Caesarean section in labour
Eight (3.4%) women in the no-scan group were suspected of intrauterine fetal growth restriction based on an estimated fetal weight below the third centile, compared with three (1.3%) women in the scan group (P= 0.13). Low birthweight (below the third centile) was confirmed in seven of the eight in the no-scan group and in two of the three in the scan group. There was no difference in the number of abnormal serum screening results: 8/233 (4.0%) in the scan group compared with 7/230 (3.0%) in the no-scan group.
Because the study was not completed, and we were not able to demonstrate any difference in the outcomes studied, it follows that the associated costs would be the same, and the formal cost analysis was cancelled.
Only 463 of the intended 800 women were recruited to the study because of the introduction of private nuchal scanning, which became increasingly popular in our local population. Patients undergoing this screening test were not eligible for the trial—this lead to difficulties in recruitment, which reduced the power of the study. In spite of this, the sample size that was achieved is still reasonably large. It is also the first such trial in a UK population. Our estimate of treatment effect is the best estimate of the true effect of a first trimester scan in this population. The data are compatible with a first trimester scan doubling the risk of induction of labour or, equally, with the scan reducing the risk of induction of labour by almost 41%. A 41% reduction in an event only experienced by 7.5% (this is the rate experienced in the control group) of the population is a smaller effect than may be deemed to be clinically worthwhile. Although a larger study would have provided a greater precision in the estimate obtained, it is unlikely that the proposed beneficial effect of a first trimester scan exists in our population. If we assume that had the study continued to accrue patients to the specified sample size and that the effect of the intervention was likely to be similar to that obtained, then the study does not suggest that a first trimester dating scan should be adopted.
Previous trials have been performed in different populations. It is reasonable to expect that the characteristics of the population studied would alter the effect of dating scans. At one extreme, a population of in vitro fertilisation patients, where dates are certain and measurements have been used as a standard,18 would not gain any further information about gestation from a scan. At the other extreme, ultrasound may be the only way to date a pregnancy if the LMP is not known. Recording the LMP may be the cultural norm in some populations but not in others. Even if CRL is more reliable than LMP in dating a pregnancy,7 the difference between the two may not be enough to be clinically important when the menstrual history is certain, and ultrasound dates may matter more if the LMP is unreliable. A dating scan would then be more useful if a large proportion of the population did not have reliable menstrual dates. The proportion of pregnancies redated by ultrasound gives an indication of the reliability of the LMP data in a given population.
The previous trials also differ in the comparisons made. Most have compared dating scans in the second trimester with no routine scan,19–22 and only two other trials used a first trimester scan.10,11 The most recent trial also used a dating scan in addition to an anomaly scan,11 as we did. Bennett et al.11 showed a reduction induction of labour for postdates, but the number of pregnancies redated was much higher (40%) than in our study (5.6%). The uncertainty of dates in almost half of their patients is very different from that found in Oxford, and this may account for the difference in the rate of induction for prolonged pregnancy. In spite of this, they showed no change in the overall induction and spontaneous labour rates were not altered. Ewigman et al.10 also used a first trimester scan and assessed the impact on the rate of induction of labour for postdates, but no other scan was routinely offered. However, as in our study, they found no significant difference in the rate of induction of labour for postdates or in the total number of inductions between the groups. Two other randomised trials evaluating scans in the second trimester have failed to demonstrate a reduction in the rate of induction of labour in the scan groups.19,20
Crowther et al.23 reported a trial assessing the efficacy of an ultrasound scan at the first antenatal clinic visit (<17 weeks of gestation) and found that 24% of women in the ultrasound group had their EDD adjusted because of a discrepancy of ten or more days from the LMP. They found no difference in rates of spontaneous onset of labour, induction of labour or elective caesarean section between the groups but did not look at the effect of a scan on induction of labour for prolonged pregnancy.
Other studies have shown conflicting results. Waldenström et al.21 also found that a second trimester ultrasound scan reduced induction for presumed postterm (3.7–1.7%, P < 0.0001) and the overall induction rate (9.1–5.9%, P < 0.0001). Both the induction rate for postdates and the overall induction rate were much lower than that found in our study. The more interventional management style in our study may have altered the impact that an ultrasound scan had on other interventions such as induction for prolonged pregnancy. LeFevre et al.22 also showed that a second trimester scan significantly reduced the rate of induction for postdates (2.14–1.6%). However, this small reduction in postdate inductions had no discernible impact on the overall induction rate (25.1 and 24.7%).
The NICE Guideline for Routine Antenatal Care (2003) recommends that all pregnant women should be offered an early ultrasound examination to determine gestational age, abandoning Naegele's rule for all cases.13 It states that ideally this should be performed between 10 and 13 weeks (as in our study) and that this intervention would reduce the need for induction of labour after 41 weeks of gestation. However, this recommendation was taken from those trials when a routine scan (including anomaly scan) was not offered. In contrast, the NICE Guideline recommends that both a dating scan and an anomaly scan should be offered. Our findings suggest that in practice, the addition of a dating scan to a routine 20-week anomaly scan may not reduce the need for induction of labour after 41 weeks and to purchasers, performing such a scan, may add to the costs.
At present, the optimal timing of a dating scan remains unclear. Nuchal translucency scans for Down syndrome screening, however, can only be performed in the first trimester. It is uncertain whether a second scan would add value, and if so, when it should be performed. The value of such a strategy must depend on the reliability of the menstrual history.
The authors thank the General Practitioners and midwives for their cooperation referring women for the study, Professor Jonathon Morris who was involved in designing the study and the interim analysis, and Dr Jill Mollison for statistical advice. The study was funded by a grant from NHS Executive, South East Region Research and Development.