A complete audit cycle of ultrasound estimation of the date of delivery
DJR Hutchon, Consultant Obstetrician and Gynaecologist, Memorial Hospital, Darlington.

INTRODUCTION
Estimation of the date of delivery (EDD) is a fundamental requirement for fetal assessment during antenatal care and is often the first thing a pregnant woman will want to know. Naegele's rule states that 280 days should be added to the first day of the last menstrual period to obtain the EDD 1, but this method has largely been overtaken by ultrasound2. Routine ultrasound dating is offered in the majority of western countries, commonly using the biparietal diameter in the second trimester to provide a gestation and from this an EDD is generated.

The accuracy of the ultrasound EDD should be regularly subjected to quality control and audited3. The audit cycle collects data to check whether accepted standards have been met. When the standard is not met, attempts are made to adjust management to improve the performance and the audit repeated to test for improved performance. We present an audit cycle of ultrasound dating in a district general hospital, completed with the adoption of a modified ultrasound dating formula.

PATIENTS AND METHODS
The ultrasound scan EDD (generated from the BPD at the routine 19 week scan) in 669 women attending for routine antenatal care under one consultant (DJRH) between January 1995 and August 1998 at a distict general hospital was compared with the actual date of delivery (ADD) as part of an audit of ultrasound dating. Only women with a singleton pregnancy, spontaneous onset of labour and a normal healthy infant which did not require any special care were included. Women who booked after 20 weeks were not included. The ultrasound chart used for dating in the maternity unit was Hadlock et al4, as recommended by BMUS5. The formula used is

Gestation in days = 6.8954 + (0.26345 x BPD) + (0.000008771 x BPD x BPD x BPD) (BPD in mm)
Gestation in days weeks = + ( x BPD) +( x BPD x BPD) + ( x BPD x BPD x BPD)

The date of the ultrasound scan, the BPD measurement in mm, and the date of delivery were collected as part of the audit. A regression formula for the equivalent gestation using the interval from ultrasound scan to delivery and the BPD measurement, and assuming a mean gestation at delivery of 280 days, was generated on a Excel spread sheet. The regression formula generated was

Gestation in weeks = 9.1688 + (0.2125 x BPD) + (0.0003 x BPD x BPD)
Gestation in days weeks = + ( x BPD) +( x BPD x BPD) + ( x BPD x BPD x BPD)

The mean difference and standard deviation between the scan EDD using the two formulae, and the ADD were calculated. The number of women delivering within a set number of days of the EDD was calculated. The analysis was repeated using the new regression formula and a published ultrasound dating formula by Altman and Chitty6. A small (55) second cohort of women delivered between September 1998 and February 1999 were used to complete the audit cycle and the analysis repeated using the new audit regression formula, the Hadlock formula, and the Altman and Chitty formula.

RESULTS

The systematic error using the Hadlock formula was - 0.96 days and the random error was 8.6 days. When the new regression formula was used the sytematic error was - 0.23 days and the random error 8.5 days. There was also an improvement in the number of women delivering within all days from 1 to 14. All women were delivered within 35 days using the revised formula, and by 38 days using Hadlocks. Using the audit formula would also lead to a lower induction rate in the small second audit population if induction was carried out between ten and fourteen days post term. In general the Altman and Chitty formula gave poorer precision with a systematic error of 2.7 days, random error of 8.6 days.


Please note - the calculations may take several seconds!
NumberTotal days differenceSum of squaresMean days differenceSD

Accuracy
Within days % Audit formula %
original data
audit cycle data

Enter any number of days in the box (below Within days ) and press calculate button

DISCUSSION
The accuracy of ultrasound dating can be assessed in two ways. Firstly the mean gestation for each ultrasound size can be compared with a "gold standard" such as IVF pregnancy where the date if embryo replacement is known with certainty 7. Secondly the EDD as predicted by the ultrasound scan can be audited and compared with the actual date of delivery (ADD) as we have done here. It would be inappropriate to include the endpoint of all pregnancies particularly those ending in miscarriage or neonatal death from prematurity. We have therefore included only those singleton pregnacies with spontaneous onset of labour and a healthy infant which did not require any special care. Since this is the outcome we are trying to predict with an EDD, this seems the most appropriate population to select. It will result in a small bias as the criteria exclude some normal pregnancies which are induced because of the fear of the risk of postmaturity. The extent of this bias however can be estimated. There will also be a small number of pregnancies induced because of either fetal or maternal concerns, in whom the baby is prefectly healthy at birth and were it not for the induction would otherwise have been included in the audit. The numbers however, are likely to be very small.

The mean days difference between the EDD and the ADD represents systematic error, while the standard deviation (SD) represents the random error. The formula resulting form this audit has less systematic error and less random error when applied to the main audited population. The systematic error in the small audit population is higher if the the audit formula is used rather than the Hadlock formula although the lower random error of the audit formula is maintained. The numbers of women delivering within a set number of days of the EDD however is improved in both data sets when the audit formula is applied. The Altman and Chitty formula does not perform as well as the Hadlock formula in this population and this is likely to reflect minor differences in population such as ethnic minorities, age and parity. Most published charts for ultrasound dating cover a wide range of gestations8. Regression formulae are commonly given and are essential if ultrasound machines are used to compute the gestation and EDD from the ultrasound measurement. However ultrasound dating is generally restricted to a much narrower gestation window9 and if the analysis to produce the regression formula is restricted, the data here suggests that a rather more accurate prediction of the EDD using the ultrasound measurements for this restriced window is possible. The completed audit cycle has highlighted the possibility that our locally produced dating formula may be more accurate than the published standards.

INSTRUCTIONS FOR USE OF THIS DYNAMIC WEB PAPER

Readers can easily see the raw data used in this paper by pressing the "View raw data" button. Other values can be entered into the formula boxes and the effect of these changes on the analysis observed. The results of statistical analysis, which takes place on the readers' computer, can be seen by pressing the "Dynamic analysis" button and further analysis of the accuracy of prediction on pressing the "Calculate precision..." button. The script to carry out the statistical analysis is contained within the HTML code and readers wish to satisfy themselves of the validity of the analysis are welcome to view the HTML code in the usual way. With large files such as this use Wordpad to open and edit the HTML code. Some workers may wish to carry out an identical analysis of their own data and this can be easily done by replacing the data within the code. If a "cut and paste" is used the process is simple and the code contains numerous comments to ensure data is entered in the correct place and with the correct syntax. Readers should note that this page can be used off-line.

References

1. Naegele F C. Erfahrungen und Abhandlungenaus dem Gebiethe der Krankheiten des Weiblichen Geschlechtes. Nebst Grundzugen einer Methodenlehre der Geburtshulfe. Mannheim:Loeffler.1812

2. Geirsson R T Ultrasound instead of last menstrual period as the basis of gestational age assignment. Ultrasound Obstet Gynaecol 1991 1 212-219

3. Persson P.-H. Ultrasound dating of pregnancy - still controversial? Ultrasound Obstet Gynecol 1999;14:9-11

4. Hadlock F P, Deter R L, Harrist R B, and Park S K. Fetal biparietal diameter: a critical re-evaluation of the relationship to menstrual age by means of real time ultrasound. J Ultrasound Med 1982;1:97-104

5. British Medical Ultrasound Society Fetal Measurements Working Party. Clinical applications of ultrasonic fetal measurements. London British Institute of Radiology 1990 6

6. Chitty L S and Altman D G. Charts of fetal size (Appendix). In Dewbury K. and Meire H. (eds) Ultrasound in Obstetrics and Gynaecology. Edinburgh:Churchill Livingstone

7. Daya S. Accuracy of gestational age estimation by means of fetal crown-rump length measurment. Am J Obstet Gynecol 1993;168:903-8

8. Altman D G and Chitty L S. New charts for ultrasound dating of pregnancy. Ultrasound Obstet Gynecol 1997;10:174-191

9. Dubose T J. Size/Age analysis Chapter 7 Fetal Sonography. Dubose T J (ed) 1998 W B Saunders, Phildelphia

Conflicts of interest - none Visits registered by WebCounter =