Thursday, 2 August 2018

More on that 'new' study on alcohol and pregnancy

I'd posted yesterday on some new work being reported by Radio New Zealand on drinking during pregnancy.

I didn't know where that work had been published because it's the rare New Zealand media outlet that will ever link to a journal. So I went to the older Superu work with which I was familiar. The numbers in the reporting looked very similar to the old study, so I figured it was safe to look to the old study's numbers on the more detailed breakdowns of heavier and lighter drinking. There's a sharp difference between heavy drinking during pregnancy and having a drink or two per week, and the media stuff I'd seen was all on the prevalence of any drinking rather than getting into that detail.

I'd figured that the new work must have been using an updated version of the Growing Up In New Zealand data - maybe a new wave of mothers had entered the dataset.

And then the Science Media Centre pointed me to the new study, out last Friday at the New Zealand Medical Journal. The nine-author piece uses the same dataset as the Superu study. The main analysis is very similar to the Superu study. It does not cite the Superu study but rather presents itself as new work.

Both studies present the raw stats and the proportion of women falling into the different consumption buckets at the different stages of pregnancy.

Both studies run some multivariate analysis using logistic regressions to get characteristics associated with different levels of drinking at different stages of pregnancy.

Superu includes some neat transition probability matrices that the new NZMJ piece didn't.

I have some difficulty in seeing the contribution provided by the new 9-author NZMJ piece given the existence of the 3-author Superu piece of three years ago.

And the NZMJ piece by Fiona Rossen, David Newcombe, Varsha Parag, Lisa Underwood, Samantha Marsh, Sarah Berry, Cameron Grant, Susan Morton, and Chris Bullen did not cite the prior work by Superu's Jit Cheung, Jason Timmins and Craig Wright.

Some questions we might then wonder about:
  • How does it take nine authors at Auckland University to replicate part of the work undertaken by three authors at Superu three years ago?
  • Did none of the nine authors know about the prior Superu work? Are any of those authors part of the Growing Up In New Zealand team? It may matter - I'm pretty sure that access to that study's data is by application, so somebody had to have authorised Superu's access three years ago. It isn't a public dataset where it's plausible that work could be undertaken that the data provider wouldn't know about. This one's locked up
  • If none of the authors and none of the referees at the NZMJ knew that this work had already been done by Superu, what does that say about standards of that journal?
I have let the journal editor know about the problem, and to their credit they're following it up (I apparently wasn't the first to note it to them either).

I wonder what the outcome will be.

I find it remarkable that the referees chosen by a local field journal in one of their areas of specialisation (go and count the number of alcohol articles that the NZMJ publishes by the public health crowd) did not catch the prior Superu work. 

No comments:

Post a Comment