Everybody who has been a child knows that it is important where you start counting (and which nursery rhyme you choose). And everybody who has read Orwell's 1984 knows that it is important to choose the important words meticulously no matter whether you want to confuse or clarify. However, both things seem to have been forgotten in the research on dating.
Expected day of delivery (EDD), gestational age, length of pregnancy, 280 days, actual day of birth, last menstrual period (LMP), ovulation, day of conception, and day of intercourse are all part of the same story, but they are different concepts and some of them are defined in more than one way. What we are discussing in our paper1 is EDD in relation to actual day of birth. In their letter Drs Zimmermann and Wisser start their first paragraph by mixing up EDD and gestational age and by introducing ovulation. We stated that EDD based on LMP on average is too early compared with actual day of birth. They state that LMP over-estimates gestational age. Thus they seem to agree with us, even though they prefer to talk about something else. We shall therefore not discuss this issue any further here.
Next, Drs Zimmermann and Wisser state that the distribution of gestational age at birth is highly skewed to the left and that our statistical evaluation based on standard deviations is definitely not allowed. It is correct that the mentioned distribution is highly skewed and that our comparison theoretically might be misleading. However, if two distributions to be compared are both skewed and even more ‘look very much alike’, then standard deviations are not very misleading when the precisions of the two distributions are being compared. Unfortunately the authors of the original papes did not publish the skewness and kurtosis of the two distributions, but the reader is solicited to judge the similarity of the shapes of the two curves in Fig. 1 of our reference2.
Finally, Drs Zimmermann and Wisser refer to new studies, published after ‘Sundhedsstyrelsen's’ search. Fortunately, one of these papers3 plus an even larger and newer study3 support our conclusion as far as can be judged from their published statistics. Unfortunately, however, the authors of these papers mix up concepts as well and introduce arbitrary numbers (280 days) into their calculations rather than basing them on observed entities. No matter whether the studies include thousands*, tens of thousand3 or millions of pregnancies4, the size of the study cannot compensate for faulty methodology. We would like to include the results of the other new paper5 in the ongoing discussion on dating as well. However, upon inspection it turned out that the methods section is not sufficiently detailed and precise to support the conclusions of the paper. We therefore wrote to the authors for clarification but received no answer. We prefer to postpone the inclusion of the study in the ongoing discussion until the quality of the research has been verified.
Meanwhile, we suggest that the owners of large and good databases start by adjusting both the technological and the non-technological method of predicting EDD so that both methods, on average, become centred around actual day of birth before they go on to compare the precisions of the two methods; furthermore, they should describe which clinical skills were applied to codinn a ‘certain LMP’. Finally we suggest that any conclusions regarding dating should be based on a systematic, critical literature review rather than on selective citing, which may lead to bias