Sir, This trial was registered in August 2009, and participants were recruited from Jan 2007-Jan 2011. Although the planned sample size was 700, the authors don't explain why 800 were actually randomised, of whom 759 ended up being analysed, but that's a minor point. Trouble starts with the primary outcome.

According to the trial registration this was "mean birth weight centiles and ponderal indices [...] measured at 14 weeks, 28 weeks, 34 weeks, at birth and 3 months post-partum". Even assuming the prenatal time points are an error, that still leaves two outcomes measured twice, i.e. four primary outcomes. That's three too many - a trial should have one primary outcome.

In the paper the primary outcome reported was different again "raw birth weight". Birth weight centile, and ponderal index were reported as secondary outcomes. None were reported at three months.

Turning to secondary outcomes the initial registration listed: maternal weight gain in pregnancy, urinary metabolomics, cord insulin, leptin and IGF-1, and placental weight, villous and vascular development. All measured at 14, 28, and 34 weeks, at birth and 3 months post-partum. On 1 Sept 2009 maternal weight gain at the various time points became the only planned secondary outcomes.

However, in the paper, besides maternal weight gain at the four time points, another eleven secondary outcomes are listed in table 2, and six more in table 3. None are pre-specified in the trial registration document.

This is a great pity, because although this seems to be a nicely done trial, readers can never be sure that they are not reading about primary and secondary outcomes selected to support a particular point of view.

Competing interests: None declared