Join 3,501 readers in helping fund MetaFilter (Hide)


Health Risks How-to
March 30, 2014 10:02 PM   Subscribe

How to think of the risks of Autism. "As a statistically minded neuroscientist, I suggest a different approach that relies on a concept we are familiar with: relative odds. As a single common measuring stick to compare odds, I have chosen the “risk ratio,” a measure that allows the bigger picture to come into focus." A succint NYT op-ed that is also a good primer on assessing health risks in general as well as the impact of media coverage on skewing risk perception.
posted by storybored (20 comments total) 13 users marked this as a favorite

 
Start from the fact that the recorded rate of autism [spectrum disorder] is now 1 in 68, according to a report released last week by the Centers for Disease Control and Prevention.
Wow, that's a tremendous rate, about double what Wikipedia lists as the global incidence, and double what the CDC found 10 years ago. I had no idea!

I love the article's graph comparing known autism risk factors and non-risk factors to the amount of press coverage. The debunking articles are quite important, but the simple act of filling in the misinformation void with the true information seems quite effective to me.
posted by Llama-Lime at 10:23 PM on March 30


I thought it was both surprising and notable that "Conceived within 12 months after older sibling's birth" was one of the larger risk factors, considering that it's almost completely avoidable if you know about it ahead of time. Why hasn't that hit the news circuit?
posted by Mitrovarr at 11:04 PM on March 30 [2 favorites]


I thought it was both surprising and notable that "Conceived within 12 months after older sibling's birth" was one of the larger risk factors, considering that it's almost completely avoidable if you know about it ahead of time. Why hasn't that hit the news circuit?

Child spacing is already promoted for that and other health reasons.
posted by autoclavicle at 11:08 PM on March 30 [3 favorites]


That risk chart, presuming its accurate, should be mandated to appear in every newspaper article on the subject. Similar could be done for every story on cancer excetera.

Of course relative risk can be misleading. If I tell you eating red meat doubles your chance of getting cancer that sounds scary, until I mention that your chance of getting cancer was 0.1% to begin with (warning, I totally made those numbers up just then!).

I'm... surprised that the incidence of autism is as high as given in the article.
posted by Cannon Fodder at 1:24 AM on March 31 [3 favorites]


As far as the relative risks vs. vaccination go, Penn and Teller pose a highly compelling argument in 90 seconds.
posted by Rhaomi at 1:51 AM on March 31 [4 favorites]


A spectrum being what it is, autism spectrum disorder currently encompasses a variety of things, including quite a lot of variations that really would not be particularly disruptive if we lived in a society that had any sort of tolerance for people who were different from an arbitrarily determined "normal". Part of the associated problem with this is the way that the media relates any risk of autism as being immediately equivalent to the lowest-functioning cases. So, essentially, it's also the problem of "eating X doubles your chance of getting cancer" followed by "cancer is a horrific disease" without mentioning how much of that increase was in non-melanoma skin cancers.
posted by Sequence at 3:54 AM on March 31 [9 favorites]


That risk chart, presuming its accurate, should be mandated to appear in every newspaper article on the subject. Similar could be done for every story on cancer excetera.

It really really shouldn't, because the immediate and obvious interpretation is that having a father over 60 adds 100% to the chance of the child having autism, thus making it certain, and that having a mother older than 35 increases the risk to at least 30%.

Risk ratios and especially odds ratios can be really confusing, especially once they've been through the telephone-game of a lay reporter. It seems like it would be simpler and more effective for scientists talking to lay publications to just directly state "X in 1000 babies born to fathers under 60 are diagnosed with autism, while Y in 1000 born to fathers 60 or over are diagnosed with autism."
posted by ROU_Xenophobe at 6:14 AM on March 31 [4 favorites]


Communicating complex health & scientific information clearly is a huge challenge - especially when it's filtered through journalists and news editors.
posted by entropone at 6:50 AM on March 31 [2 favorites]


Seconding ROU, I teach high school seniors and they definitely don't understand "Y increases by X percent" (not even getting into when Y is a probability and not a number).
posted by subdee at 8:41 AM on March 31


People also like to blame complex things that are beyond their control for illness - I just got back from a week long SCA thing where we had an outbreak of what seems to have been (based on onset, symptoms, recovery and spread) norovirus. It was interesting how many people insisted that it was the campground's water system and something really exotic.
posted by Kid Charlemagne at 10:13 AM on March 31 [2 favorites]


Of course relative risk can be misleading. If I tell you eating red meat doubles your chance of getting cancer that sounds scary, until I mention that your chance of getting cancer was 0.1% to begin with (warning, I totally made those numbers up just then!).

Since these are all rate ratios for autism, the comparisons don't suffer from that problem.
posted by Mental Wimp at 11:52 AM on March 31


subdee: "Seconding ROU, I teach high school seniors and they definitely don't understand "Y increases by X percent" (not even getting into when Y is a probability and not a number)."

Dead quote from a college-educated, intelligent ex-girlfriend: "The doctors said I only had a 10% chance of ever conceiving, so we quit using birth control - and so I got pregnant when we really couldn't afford it."

10% <> 0%. And I swear, if you spoke to her a few minutes or an hour, you'd think she possessed above-average intelligence.

The biggest problem with communicating risks and probabilities to the general masses is that they are actually unimaginably stupid en masse, and will hurt themselves with those pointy facts like a chimp with a handful of razorblades and needles.
posted by IAmBroom at 11:53 AM on March 31


"Sibling is a science or engineering major in college" -- I think that people on the milder end of the spectrum are often attracted to these fields, so might this not just reflect that people with autism spectrum disorders (who are more concentrated on these fields than in the general population) are more likely to have siblings who also have autism spectrum disorders?

I wonder whether the risk for non-verbal autism is actually increased by a factor of two, or do most of the people with the sciency siblings actually get diagnosed with a higher-functioning form of the disorder?

For that matter, I have the same question about some of these other statistics. Does the risk of severe low-functioning autism really scale the same as the risk of Asperbergers/high functioning autism in all of these cases?
posted by OnceUponATime at 12:54 PM on March 31 [1 favorite]


The terminology is a bit sloppy in the article, both with regard to autism/spectrum and risk/risk factor. This article has a table of risk factors, which describe correlations rather than causations. Using the term 'risk' alone almost implies that by changing that factor for a person it will change the chance of that person having ASD, but that's not the case.

So if something is not a risk factor, it's safe to ignore it as a potential cause. But if it is a risk factor, more information is needed before causation can be established. It looks like there's quite a bit of research into time between pregnancies, so that is probably a true risk as well as a risk factor, but I don't think it seems likelythat discouraging siblings from STEM careers will stem ASD prevalence.

I think many people who are aware of the distinction would read the article and make the correct substitutions in their head because they'd be making assumptions about what could be and what actually has been studied, but it could be very deceptive to an unprepared audience. I still liked and appreciated the article even with its current terminology misuse.
posted by Llama-Lime at 1:21 PM on March 31 [1 favorite]


This article has a table of risk factors, which describe correlations rather than causations.

I think it is more correct to say "correlations, but not necessarily causations." But there is no official consensus on what "risk factor" means. CDC uses one definition ("risk factor: an aspect of personal behavior or lifestyle, an environmental exposure, or a hereditary characteristic that is associated with an increase in the occurrence of a particular disease, injury, or other health condition."), WHO uses another ("A risk factor is any attribute, characteristic or exposure of an individual that increases the likelihood of developing a disease or injury."). However, a quick scan of the epidemiologic literature shows that by usage it usually refers to factors associated with the outcome of interest. Causal factors are modifiable characteristics that change the probability of disease.
posted by Mental Wimp at 1:32 PM on March 31 [1 favorite]


It's not that I think there's likely to be confusion about whether the "science and engineering" one is likely to be correlation or causation -- as you say it doesn't seem likely that discouraging siblings from STEM careers will help, and probably few people would think it would.

But I wonder whether it's a really simple correlation -- just the probability that you will have autism given a sibling with autism, times the probability that someone who is in science and engineering will have autism? Or does that not predict the risk increase? I suppose there are probably a fair number of undiagnosed but affected people in science and engineering too (again, on the milder end of the spectrum), and if we could somehow include them, would that explain the correlation?

Or is it really that autism and being interested in science (but not autistic) are related in some way? Do the genes that cause autism somehow also express themselves as interest in science instead, sometimes? It's weird to me to think that would be genetic, and the idea kind of disturbs me. But maybe introversion is genetic and linked to the genes for autism, and science is a good field for introverts?

And as I say, I'm wondering about the severity. If it is some kind of genetic link, whether of the trivial kind or the "genetically predisposed to be a scientist" kind, I would expect that being related to people who are high-functioning enough to get degrees in STEM would be correlated with relatively high-functioning autism among the siblings as well. Or does it not work that way?
posted by OnceUponATime at 2:16 PM on March 31


I'd like to know how much of this is linked to funding for autism. In many parts of Canada, you can get anywhere from a few thousand to $40,000 annually to address the needs of a child with autism. So, for example, some people will answer certain assessment questions so they can get the funding. If you have some other disorder, you don't get anything to help pay for speech therapy, play therapy, tutoring, etc. So some people look for doctors who are willing to make the assessment and say it is mild, just over the side of being borderline, etc. I have no idea if this is the case in other countries. But I know it works here. I was sad to find out my child's needs are not covered by funding, because it's not autism. I know people whose kids have all sorts of challenges, but only autism comes with $500 a month through age 18 for these supports. And autism comes with access to an aide at school - another incentive to get a diagnosis. Is it possible this influences whether people pursue a diagnosis now?
posted by Chaussette and the Pussy Cats at 5:10 PM on March 31 [4 favorites]


Couldn't the time between pregnancies be due to the fact that women who are e.g. older, or less fertile for other reasons, or have health issues might be more likely to have a bunch of kids close together while they still can? So it could be those other issues that are linked to the autism risk, rather than the pregnancy spacing itself.
posted by lollusc at 6:02 PM on March 31


Couldn't the time between pregnancies be due to the fact that women who are e.g. older, or less fertile for other reasons, or have health issues might be more likely to have a bunch of kids close together while they still can? So it could be those other issues that are linked to the autism risk, rather than the pregnancy spacing itself.

That's possible. Someone who has issues with fertility might be more likely to have multiple children at once due to fertility treatment, which brings it back to the first criteria of having a twin with autism elevating the risk that the other twin will have it as well.

Income is likely related to child-spacing. We know that access to birth control and abortion is limited for lower-income people. If you're not able to use birth control, you're more likely to have children.
posted by autoclavicle at 11:45 PM on March 31


lollusc: Couldn't the time between pregnancies be due to the fact that women who are e.g. older, or less fertile for other reasons, or have health issues might be more likely to have a bunch of kids close together while they still can? So it could be those other issues that are linked to the autism risk, rather than the pregnancy spacing itself.

Even assuming the studies involved didn't correct for age, it would be hard to explain how insufficient child spacing so massively outweighed the risk conveyed by maternal age, if age was the cause of the risk due to child spacing. Plus, older mothers often have more trouble getting pregnant, which to me suggests they'd be less likely to have children close together.

On the other hand, having children close together is already associated with various pregnancy complications, and I can easily believe the increased risk of autism is due to those complications. In particular, it's associated with premature birth, which is directly associated with autism. It'd be interesting to know if the study that correlated insufficient child spacing with autism corrected for prematurity.

I'm still surprised by the strength of the insuffucient child spacing -- autism link. Seriously, it more than doubles the risk - that's a big deal. That puts 'having two children one right after the other' in the same catagory as 'smoking and drinking while you're pregnant' and 'eating thermometers'. They should have PSAs about this.
posted by Mitrovarr at 1:53 PM on April 1


« Older “Just as leaves in a tree are naturally arranged t...  |  If you don't like Marcellus La... Newer »


This thread has been archived and is closed to new comments