More BJOG salami

Last year we drew readers attention to a florid example of salami publication in BJOG (click here).  A group of authors from Denmark had taken five papers over 56 pages, to report a single study of the relation between low to moderate maternal alcohol consumption in pregnancy and child development at five years. Each paper was based on the same sample of 1628 women, reported the same exposure variables, adjusted for slightly different but overlapping covariates, and reported slightly different but overlapping outcomes. All five papers concluded that there was no proven harm from low to moderate consumption, but that mothers should play safe and not drink at all. The BJOG added it’s own two slices of commentary which concluded the same thing.

In the latest issue of the same journal (click here or here skogerbo bjog 2013) the same group have added yet another eight page slice about the same population and exposure variables, another overlapping set of covariates and a new five year outcome, the Strengths and Difficulties Questionnaire (SDQ) and its various subscores. The findings were the same, namely:

“This study observed no consistent effects of low to moderate alcohol consumption or binge drinking in early pregnancy on offspring behaviour at the age of 5 years.”

BJOG then added yet another two slices of commentary (here and here) each with the same conclusion again.

Last time (click here) I noted that the authors had mentioned eight other five-year outcomes, and jokeingly wondered if we could expect eight more slices of salami!  It looks like my fears were justified; this paper reports just one of those eight other outcomes! Seven to go!

To summarise; one study has now been sliced into six separate reports covering in total 64 pages of journal, and the journal has added four separate commentaries. Every single one of which says what everyone knew already, namely that low to moderate maternal alcohol consumption has not be proven harmful to the baby, but that to play safe mothers should abstain altogether!

This sort of nonsense is wasteful, and makes the literature difficult to interpret. The Committee on Publication Ethics (COPE) (click here) is quite specific that it is scientific misconduct: “splitting up papers by outcomes is not legitimate”. However, it is a rather mild form of misconduct, so referral of the authors to their university bosses seems rather harsh. Perhaps public shaming will have some effect.

If they do it again I’m going to tweet them round the world!

Jim Thornton