Lancet asked to review controversial birth study

The Lancet’s  Ombudsman, Dr Charles Warlow, has ruled that criticisms made here of a WHO study of the risks of elective Caesareans and published in the journal raise “considerable concern”.
 
The study, carried out in Asia, claimed that women who choose to have a Caesarean without medical reasons are nearly three times as likely to die or suffer complications as those who have a normal vaginal delivery.
 
The result depends entirely on a statistical correction for “confounding factors” – the differing baseline risks of the individual women - since the raw data show elective Caesareans are actually safer. I described this result as extraordinary, but when Penny Christensen of Birth Trauma Canada sent my analysis to The Lancet, together with some criticisms of her own, it was dismissed because it had not been peer-reviewed.
 
Dr Warlow is more straightforward. In a letter to Ms Christensen, he says I was right to point out “simple arithmetic errors” in the paper which do not give him confidence that the more complex mathematics are correct, in particular the adjustment of risk. He has asked The Lancet to review the paper again with the help of a statistician.
 
 “It is exceptionally unusual for an adjustment like this to make a qualitative rather than just a quantitative difference to an odds ratio, here from less risk of antepartum caesarean section without indications to more risk ... moreover Hawkes is right to point out that the adjustment is of a proportion based on only 9 events which must be shaky to say the least. (His italics)
 
“This clearly needs sorting out by someone more statistically competent than me and I will ask The Lancet to do this.”
 
Dr Warlow dismissed other objections raised by Ms Christensen, saying that in publishing the paper The Lancet did not stray outside its usual procedures and practices, and that the paper had been peer-reviewed by at least three clinical/scientific reviewers and one statistical/epidemiological reviewer. “We can really expect them to do more than they actually did with this paper” he said. He also dismisses the charge that The Lancet censored criticism of the paper.
 
He says that one criticism was published (noted here) and that three other critical letters were submitted, but only one made reference to the statistical issues – “but without the sort of important detail that Hawkes provided.”
 
We will await Lancet’s response.