SEARCH

SEARCH BY CITATION

Introduction

  1. Top of page
  2. Introduction
  3. Prenatal screening
  4. Fetal viability
  5. Prolonged pregnancy
  6. Fetal growth assessment
  7. References

Dating by ultrasound in the first half of pregnancy has become a routine part of antenatal care in many units around the world. A recent commentary in this journal1 questioned the accuracy of ultrasound and argued for the use of menstrual history with some modifications of Nagele's rule. This would be a regrettable and retrograde step. Recent research has shown conclusively that, where equipment and trained personnel are available, ultrasound is the method of choice for dating pregnancies. We present a review of these findings and put their significance into a clinical context.

In comparing the measurement of gestational age by menstrual dates and by ultrasound, one has to look at systematic and random error. In each case, ultrasound performs considerably better.

Systematic error states how accurately, on average, the measurement reflects the truth in the sampled population; this is important for definitions and policies based upon them, such as the timing of induction of labour for post-term pregnancy. Nagele's rule uses the mean length and predicts this to be 280 days2. Over 160 years later it is surprising how accurately this figure still reflects the mean length of pregnancy, including spontaneous and elective deliveries, in a database of 25,000 pregnancies with ‘certain’ menstrual dates: 279.7 days3. But in view of the negative skewness of gestational age distributions, and the possible influence of induction policies which may vary in different populations, the mean or even the median is not the best parameter to derive the ‘typical’ length of pregnancy. Instead, in large enough databases this can be derived from a modal value, found to be 283 days3. This is the same as that in the study by Bergsjo et al.4 and which Olsen and Clausen’ now propose as the ‘true’ length of gestation.

In contrast, using routine second trimester ultrasound the modal length of gestation was found to be 281 days in two recent large studies from different populations3–5 using different dating formulae6,7. The accuracy of these formulae were tested in studies of assisted conception pregnancies, and their systematic error was in each case found to be less than a day8,9. A study looking at the basal body temperature shift in 1408 conception cycles10 also concluded that the average gestation of term pregnancies was 281 days. Therefore, the typical length of gestation is not 283 days. Olsen and Clausen suggest that the number of post term pregnancies could be reduced by using 283 days, but conclusions about the length of pregnancy cannot be derived from menstrual data as the lengths of the follicular phase before conception are not known.

Random error assesses how consistently the measure is performing within the population. The problem with reliance on menstrual history for dating pregnancies is well known. Firstly, many women do not remember or are uncertain of the date of their last period11–12. Secondly, even if the menstrual dates are considered ‘certain’ or ‘reliable’, they have a much wider distribution which tends to be skewed towards an over-estimation of gestational age when compared with ultrasound5,11,13–17. This variation is again related to the fact that in many instances conception occurs not in mid-cycle but later10,8,9. In a study of basal body temperature10, the day of the temperature shift within the cycle had a wide distribution and positive skew, on average 16.5 days after the first day of the last menstrual period; whereas the interval between temperature shift and birth had a narrow and near-normal distribution, averaging 264.8 days. In a biochemical study of 75 ovulatory cycles, the average LH surge was also noted to be 16.4 days into the cycle19. Significantly, this study also showed that in half of the cycles in which ovulation was delayed until after day 18, this delay would not have been predicted on the basis of the length of the menstrual cycle.

In contrast ultrasound error is normally distributed and has a much narrower error margin. Studies in assisted conception pregnancies showed that in single units, the random error with a formula for biparietal diameter and femur length can be as low as 2.47 and 2.9 days SD9. Head circumference as a single parameter has been recently reported to have a SD of 3.8 days6. Even in routine ultrasound clinics across 25 different units, dating by biparietal diameter still only had an error of 4.3 days SD8. Ultrasound is better at predicting the actual date of delivery than the ‘certain’ last menstrual period alone or in combination with menstrual dates, even when the discrepancy between the two measurements is less than seven days3,5.

Although ultrasound dating formulae have been derived from pregnancies dated by last menstrual period, they are more accurate in day-to-day use than the last menstrual period itself The formulae are based on relatively few pregnancies with carefully checked dates and regular, 28 day cycles, and any small variation is eliminated by regression of data to derive an average formula. Individual variation in size (eg, due to fetal sex or ethnic group) may affect the measurement, but this has a relatively small effect. The first half of pregnancy is a period of rapid cell division, and gestational age is by far the strongest variable affecting fetal size13. Even singletons and twins, who exhibit considerable differences in the third trimester and at birth, are virtually indistinguishable in size in the second trimester, such that the same dating formulae can be used20. The accuracy of the formulae is also confirmed in studies based on assisted conception pregnancies8. By convention, the formula is expressed not as conceptual age but as menstrual (gestational) age after adding a standard 14 days; thus the length of pregnancy relies on fetal biometry dating and not also on an unknown component (ie, the length of the follicular phase). This is a principal reason why ultrasound dating formulae perform better in day-to-day use.

The accuracy of ultrasound dating is especially important at the extremes of pregnan3,13,21,22. The significance of the better performance of scan dating compared to menstrual dating becomes clear when we consider some clinical implications. They show how both systematic and random errors can have substantial effects.

Prenatal screening

  1. Top of page
  2. Introduction
  3. Prenatal screening
  4. Fetal viability
  5. Prolonged pregnancy
  6. Fetal growth assessment
  7. References

Because of the close relation between gestational age and normal limits of maternal serum alpha-fetopro-tein and human chorionic gonadotrophin, the results of screening tests for chromosomal abnormalities are heavily dependent on precise gestational age, especially when serum levels are in a borderline range23. Menstrual date error will result in many false results. Often, as the last menstrual period error tends. towards overestimating the true gestation, the risk for the screening test is falsely magnified, which leads to unnecessary referrals, investigations and maternal anxiety. The sensitivity and specificity of maternal serum screening is improved if ultrasound dates are used23,24.

Fetal viability

  1. Top of page
  2. Introduction
  3. Prenatal screening
  4. Fetal viability
  5. Prolonged pregnancy
  6. Fetal growth assessment
  7. References

Menstrual dates systematically underestimate the prevalence of preterm deliveries25,26. At very early gestations menstrual dates also tend to underestimate fetal age compared to the early scan22. For example, bleeding in early pregnancy, a strong risk factor for subsequent preterm delivery, may have been mistaken for the last menstrual period. The curve of fetal survival rises steeply between 24 and 28 weeks, from about 20% to 80%27. Hence at these early gestations, even a few days of variation in either direction will have an impact on the assessment of a baby's chances, on the advice given to parents, and on clinical management decisions.

Prolonged pregnancy

  1. Top of page
  2. Introduction
  3. Prenatal screening
  4. Fetal viability
  5. Prolonged pregnancy
  6. Fetal growth assessment
  7. References

From 42 weeks gestation by last menstrual period it is increasingly likely that the dates are wrong3. The discrepancy with early scan dates exceeds the 95% confidence interval of routine ultrasound dating error. Altogether, 72% of pregnancies presumed to be post term (>294 days), according to menstrual dates, are not post term according to scan3. Extremely few pregnancies extend beyond 285 days from the time of ovulation10,28, or beyond a gestational age of 300 days (42 weeks 6 days) by scan dates, in populations where there is no routine induction policy for post term pregnancy17. Although dates are usually ‘confirmed’ by scan, they are often not adjusted unless the discrepancy with the scan is at least 7, 10 or 14 days; yet compared with all the different combinations, the actual date of delivery is best predicted by scan alone16, and the rate of misdated inductions would be minimised if the last menstrual date is ignored altogether.3.

Fetal growth assessment

  1. Top of page
  2. Introduction
  3. Prenatal screening
  4. Fetal viability
  5. Prolonged pregnancy
  6. Fetal growth assessment
  7. References

The assessment of weight-for-gestation depends on accurate dates, and is important for prospective management as well as retrospective assessment. Some small for gestational age fetuses have diminished fetoplacental reserve; in the preterm period they have an increased rate of adverse clinical outcome compared to fetuses at matched gestations which are not small for gestational age29. The preterm weight curve is steep and as the weights are lower, a few days variation in gestational age will result in a much larger error as proportion of actual weight, and a much larger effect on whether that weight is considered to be within or outside normal limits.

At post term, small size for gestation is the single largest risk factor for perinatal mortality30. Yet growth failure may be easily missed. The systematic over-estimation of gestational age when calculated by menstrual dates results in an artificial flattening of standard birthweight curves at term, which is apparent even in recent large scale studies31,32. In contrast, the weight increment continues after 40 weeks if early ultrasound dating is used33,34,35,36. This also reflects the true pattern of intrauterine weight gain as shown in ultrasound fetal weight studies37–39. The implication here is that real growth failure at term and post term would not be recognised if the charts expect no growth. It is possible that the oft- quoted fetal risks associated with the ‘post-maturity syndrorme’40 are not really a result of postmaturity, but unrecognised placental failure at term.

In summary, accurate dating is a linchpin for good pregnancy care. There are many problems with dates based on menstrual history, which result in significant systematic errors as well as large random variation. There is now conclusive evidence that scan dates are more accurate than menstrual dates. Routine ultrasound has resulted in the ability to better define the normal length of pregnancy and has given us a tool for improved assessment of fetal growth, prematurity and prolonged pregnancy. Ultrasound scan dates ought to be used not just to correct menstrual dates if there is a certain discrepancy, but in preference to menstrual dates, as the first choice for dating all pregnancies.

References

  1. Top of page
  2. Introduction
  3. Prenatal screening
  4. Fetal viability
  5. Prolonged pregnancy
  6. Fetal growth assessment
  7. References