Can I start a pedant's thread about statistics?

[peanutbutterkid]

Coz I'm gonna do it anyway...

I am dying to moan about a headline in the yesterday's Guardian. Front page said that "146 healthy babies lost every year due to poor Down's test". If someone else has read the article, tell me if I got this wrong, but I think not.

In the text of the article it says that in 2008, 14,460 women had amnio due to high risk of Down's syndrome. 1112 of these women had terminations after results showed baby had a chromosone problem (1/14 of the women who had amnio). In addition, 146 "healthy" babies were miscarried after amnio (about 1%, which is what I've read everywhere as the national figure for m/c after amnio).

58 of the 146 mothers might not have had a high risk stat result for Down's Syndrome, if a better quality screening test were available. So they wouldn't have had amnio and then m/c'd. This is the gist of the article. Quoting: "...58 of the 146 healthy babies met avoidable deaths annually."

So why did the headline say that 146 healthy babies had been lost due to poor quality testing??? Why not say 58?

Plus, does the 146 figure include any babies with chromosone problems -- surely it must, else chromosone-affected pregnancies never m/c after amnio? We would expect 1/14 of the babies subjected to amnio to have a chromosone problem, based on the figure for terminations, so 1/14 of the 58 (or 4 babies) may have had a chromosone problem, anyway.

So the better quality test would lead to between 54 and 58 fewer miscarriages of "healthy" babies. 54 or 58 is what the headline should have said.

This is an emotive subject, but how can we discuss it properly when it gets badly reported? And on the front page of a quality broadsheet, what are their fact checkers paid for???

You're right, 'due to' means that the number of babies they are referring to must be lost because the test is bad, and would not be lost if the test were better. So it should be the 58.

Presumably once the amnio has been done they can tell which babies have chromosomal abnormalities and which don't, so your second point depends on whether they are counting total number of MCs after amnio, or whether they are actually looking at the MCs after amnio, with the knowledge of which are 'healthy', and only counting those.

They may be assuming that if you agree to the amnio in the first place (knowing the risk of MC) that you have already decided to terminate if the test is positive, in which case there would be only healthy babies left to miscarry.

Also, leaving aside the stats and bringing it back to grammar, it shouldn't have been "due to", it should have been "owing to". Due to means caused by; owing to means because of. When you see due to, it's more often than not wrongly used.

Good point, and one that I wasn't aware of.

Definitely should have a corner for stats, they are very, very badly reported / understood / used. On The Apprentice last night they kept banging on about how "only 2.2% of women have a home birth" as a reason for not selling birthing pools at the Baby Show. But that's 2.2% of the general population, not of the visitors to the show which is likely to be far more skewed towards homebirthers. Tsk.