Olsen O. et al.1 (Vol 104, November 1997) stated that routine dating has not been shown to be more accurate than the calendar method to predict the day of delivery based on the fact that last menstrual period (LMP)-dated and scan-dated pregnancies only differed by three days with an identical standard deviation. Whilst it is well known that LMP does over-estimate gestational age because late ovulation is much more likely to occur than early ovulation, the problem of the dating story lies in the distribution of gestational age.

Distribution of gestational age at birth is highly skewed to the left. This is due to the fact that deliveries do occur 26 weeks after ovulation, but they never do 45 weeks after ovulation. Even in so called ‘term’ pregnancies (37 to 42 weeks) the distribution is not Gaussian distributed, as the authors themselves have found different values for the mean, median and mode. As a consequence, comparing the standard deviation between LMP-dated and scan-dated length of gestations is definitely not allowed, since the standard deviation is strongly influenced by the lower tail of the distribution.

The proper way of statistical evaluation is to compare the proportion of deliveries occurring beyond 14 days after the median length of gestation. By doing so, Mongelli et al.2 found a reduction of post-term pregnancies from 11.5 to 3.5% in 34.249 pregnancies and Tunón et al.3 reduction from 10 to 4% in 15.000 deliveries, both sample sizes much higher than the ‘only’ study on this topic including 1650 deliveries mentioned by the authors As stated by Backe et al* routine ultrasound predicted the day of delivery more accurately than the calendar method. As a consequence the MIDRS leaflet is based on the best available evidence both by randomised and population based studies


  1. Top of page
  2. References