Sunday, 21 August 2016

A Mum's Guide To Dodgy Science: Cause v Correlation


Time for another in my (very) occasional series of posts giving you tips on how to spot a dubious science story. The previous posts covered Publication By Press Release and the problem of Comparing People To Petri Dishes. This one has a few less P's in it but it's a biggy, something that crops up time and again in the media: causation v's correlation.

I've touched on it before (in this post), but the problem is essentially this: if a study finds that, say, people who eat lots of broccoli are better at maths than those who don't, then it could be claimed that eating broccoli causes super maths skills. 

But is the broccoli really the cause? It could just be a coincidence or there could be another factor that causes an increase in both maths geekiness and tiny green tree consumption. 

If you fancy making up some correlations of your own have a play on this website, apparently, increasing sour cream consumption results in more Lawyers in New York.

This might sound quite obvious (not the sour cream bit) but it's all too easy to fall for it. Especaially if the cause and effect seems to back up our own beliefs. This is a whole other problem (known as confirmation bias) and it's really really hard to avoid. Show me a headline that says getting a science degree causes you to be a totally awesome person and a little bit of me would be very happy to believe it (ok a pretty big bit). The moral of the story is, if you disagree with the findings of some research, be skeptical about it. If you agree with the findings be really, really skeptical about it.

So if we accept we are all just human, how can we spot when there is a causation v correlation issue in an article? 

There are some key phrases you can look out for, things like "linked to" "associated with" "relationship between" . Basically anything that suggests a connection between two things but doesn't explicitly say "causes" is a bit of a red flag.




*It's usually the mothers right?

As ever I tend to get around to writing these posts because I saw something that annoyed me, so here's that something:


There is a strong positive relationship between planned birth at home and breastfeeding rates, according to researchers.
Their study review found breastfeeding was twice as likely among mothers in UK and Ireland who had a planned home birth, compared to hospital births.
The article goes on to list a number of reasons why this may be: Home birth mums are looked after by midwives not doctors, they aren't confused by lots of different medical professionals and are less likely to have medical interventions or pain relieving drugs. They might also be more likely to have skin to skin contact immediately after birth and less likely to have formula on hand. 
All of these things are mentioned in the paper the article is based on and they may well all have contributed to the results. BUT there is also a glaring correlation issue:
In the UK and Ireland very few births happen at home. In the two populations studied it was just 1% and 2%. So no one was going along with a home birth because it's just what everyone does. It's a carefully considered and sometimes fought for decision. Hospitals deal with everyone from super healthy twenty somethings to those with multiple health and social issues but home birth mums tend to be similar. They are usually educated, healthy and relatively well off. Exactly the same demographic that is most likely to breast feed, wherever they give birth. 
This issue is discussed at length in the paper, it's such a biggy that the authors spent time on complex stats to try to adjust for things like socio economic status and if the mother had a live in partner. Yet the article doesn't mention this problem at all. 
The paper is also quite clear that, even with those statistical adjustments (and they are never perfect) there is still the issue of belief. 

Those mothers making an active choice to give birth at home do so because they believe it is best for them and their baby. They are fairly confident that birth is a normal, natural event which they are perfectly capable of going through with no need for artificial intervention from doctors and modern medicine. It would be very odd then, for women with that belief not to extend it to the normal and natural act of breastfeeding. 

Sadly belief and determination are not enough to make breastfeeding successful, I write this as someone who had both and had a hell of a time with my fist baby. But they do help, sometimes a lot. Yet the article makes no mention of the importance and power of women's beliefs.

Instead, the story painted is one that fits neatly into the current ideal of childbirth: Doctors, hospitals and pain relief are bad. Midwives, breastfeeding and all things natural are good. 

But to get back to the causation issue - why does this matter?

This article didn't appear in a tabloid paper. It was in a nursing magazine. Most people reading it would expect it to be an accurate source of information and they probably don't have the time to dig through the original research to check on that. Yet the article follows the same formula we see in the general media. It cherry picks the bits of the paper which will most appeal to it's audience but leaves out some very important problems.

We all love a quick fix. Take this pill to get slim, eat today's favourite "super food" and stop feeling tired all the time. But it is rarely that easy. Discouraging doctors, epidurals and formula in the hospital would all be doable and pleasing graphs could be produced of their decline. But it could mean coercing women into births that are more painful or risky than they would otherwise choose. If these things aren't even the main cause of the reduced breast feeding rate, if hospital birth over all is more a correlation than a cause then just jumping into the quick fix could do far more harm than good.

SBx

PS. There are a bunch of other issues with article, (including an error in the first paragraph), there are also some weaknesses in the paper but for brevity I've resisted a full rant!






No comments:

Post a Comment