I was going to write about a recent study that purports to claim that smoking pot causes schizophrenia that's been making the rounds lately. Unfortunately (or fortunately, depending on how you see it), this week's host of the Skeptics' Circle, Mark Hoofnagle, beat me to it.
Can you say, "Correlation does not necessarily equal causation"?
The reporters who hyped the study and the investigators who enabled them should repeat this 100 times. Maybe it'll sink in.
More like this
This week, the 66th Meeting of the Skeptics' Circle just so happens to coincide with the 66th Meeting of the International Society of Skeptics; so they decided to hold a joint meeting. This week's host, Mark Hoofnagle, has the abstract book all lined up for you to peruse. Excellent stuff. If only…
Well, it's frigid indeed this weekend around Castle Orac. I need something to warm the cockles of the circuits in the deepest, darkest depths of winter, and I know just the thing: A heapin' helpin' of the best skeptical blogging out there. Fortunately, the next meeting of the Skeptics' Circle is…
This has been a bad week at skeptics' school. Apparently, skeptical bloggers have been misbehaving left and right. Apparently we as the skeptical blogosphere have been very, very naughty indeed. Worse, the essays that we've handed in are apparently not pleasing to the teacher. Worst of all, we've…
The next Meeting of the Skeptics' Circle is fast approaching, and, if you haven't already done so, if you're a blogger you should be thinking about which one of your pièce de résistances of skepticism you want to submit. Guidelines are here. Next week's host, Dad of Cameron at Autism Street has…
I've been wondering about the effects of non-standard types of second-hand smoke.
Can you say, "Correlation does not necessarily equal causation"?
The reporters who hyped the study and the investigators who enabled them should repeat this 100 times.
Never mind the reporters who hyped this study, how about just all reporters who write about science in general? Not one seems to actually understand this concept.
I'm beginning to think 'correlation != causation" ought to be one of the foundation statements of our educational system. So few people get it; it's so critical.
I think it's worse than correlation!=causation problems. Legal drugs like alcohol and cigarettes have a higher correlation to psychosis than marijuana, but you don't hear them maligned.
I think it's correlation!=causation in addition to puritanical fearmongering.
Joseph Hertzlinger:
If my experience in Amsterdam was at all representative, the smell of burning cannabis doesn't stick to clothes nearly as badly as tobacco smoke does. This is an odd phenomenon which I'm rather at a loss to explain.
Obviously, secondhand cannabis smoke causes memory loss, even in textiles.
Once again, Orac's summary misses the point. Ignoring the paper and shouting "Correlation does not equal causation" is only an acceptable practice on Slashdot. There are several paragraphs in the paper on this issue, and I would have preferred to hear someone's commentary on them. You have failed your blogroll probation--goodbye.
Don't let the door hit you in the posterior on the way out, blogospherically speaking; you've failed your commenter probation.
Glad to see you've decided to mention this one.
As one who has used marijuana to self medicate in the past (actually still occasionally supplement my dimenhydramine with it), I have a hard time believing that the correlation isn't the result of the fact that people with a variety of neurological disorders use a variety of drugs to self-medicate. Many of us, because we lack any real access to healthcare.
I commented elsewhere on this article:
Instead of assuming that the chain of causation goes 'marijuana use -> propensity to become schizophrenic', why not assume 'propensity to become schizophrenic -> marijuana use'? This would make a lot of sense mechanistically, since people in psychological pain are well known to medicate themselves.
And for all we know, this self-medication with cannabis is effective, and some people who would otherwise have become shizophrenic do not.
ZOMG! MARIJUANA PREVENTS SCHIZOPHRENIA!
Why didn't they address the issue of self-medicating? Please. They've taken too many good Rx meds away and made dispensing of many of those that remain too tedious to supervise and prescribe.
Blake -
Simple reason I found for this was the use of Cigarette vs. hand rolling tobacco - I don't know what they put in cigarettes, but they small a lot stronger, and the smell persists much more.
Actually, in Europe, the cannabis-schizophrenia case is used as an example in the introductionary course of Epidemiology as a prime example of how "Correlation does not necessarily equal causation". Maybe I am deluded (and biased, as I am living in Amsterdam, the cesspit of sin and marijuana) but I thought that the discussion on cannabis/schizophrenia was a really ancient one...
The investigators of the cannabis-schizophrenia correlation have tried very hard to eliminate confounding effects. They point out, for example, that the marijuana use precedes the development of symptoms, and that there is a suggestion of a dose-response relationship. There are even plausible biological scenarios as to how the effect could occur, in that there is evidence of a deficiency of an enzyme that makes the neurotransmitter GABA in schizophrenics, and cannabinoids act on GABA terminals to inhibit GABA release, suggesting that they could exacerbate a flaw in GABA mediated synaptic transmission.
I remain somewhat skeptical. The problem is that there is substantial evidence of a genetic link in schizophrenia, which implies that people who develop schizophrenia are abnormal in some way even before the emergence of symptoms. And since symptoms typically emerge in early adulthood, researchers who want to exclude the possibility that of reverse causation (e.g. self-medication by people with schizophrenia) are limited to looking at people who use cannabis in their teens. The correlations are thus strongly influenced by an excess of individuals who are very heavy cannabis users in their early teens and subsequently develop schizophrenia. But heavy cannabis use in the teenage years is fairly unusual, and could suggest an underlying pathology.
The authors acknowledge that schizophrenia incidence trends have not followed cannabis use, but argue that there are enough uncertainties that this could be missed. This is plausible because the effect size is not huge. Based on the average risk and cannabis uses statistics, they extrapolate that perhaps 14% of psychosis might be cannabis-related (which as they acknowledge, ignores multiple uncertainties). Keep in mind that the population incidence of schizophrenia is around 1% or so, depending upon the diagnosis criteria used.
So while I think that the link very well could turn out to be real, it does seem like it has been oversold. It's a bit hard to scare people with a more balanced statement, like "If you smoke pot every day, there is a 0.14% chance that it will make you go crazy" or "If you have a family history of schizophrenia, you probably shouldn't smoke a lot of pot when you're a teenager."
The only good thing about this type of article is that it gives you more opportunities to introduce the concept of correlation not equalling causation. My 14 year old had a bunch of friends over, and they were reading about the study that purports to prove that even diet sodas cause obesity. "Correlation does not equal causation!" I said to them. They agreed that they didn't know what that meant, but it sounded really good. So I explained to them that people with poor eating habits will often drink diet soda in large quantities, so it's the bad eating habits, not the soda. They understood, we had some more chat about it, and another seed of rational thought has taken root in young minds. I do what I can. . .
(BTW, this is also somewhat germane to the ADD discussions going on at Pharyngula and Retrospectacle, where mention has been made about ADD and self-medication. The ADD comes before the self-medication, too.)
I'm sure a correlation study would find (or has found) that drinking alcohol is related to depression. It would be far-fetched to claim alcohol causes depression, wouldn't it?
I don't think correlation studies are useless. But it's important to understand confounds. Most studies are probably wrong. What matters is the scientific process that comes after the initial finding.
Uh...why?
I take that example back. Alcohol can be a depressant.
A better example might be Parkinson's inversely correlated to smoking.
Example of what? This has been a fairly consistent observation, and a protective effect of nicotine is biologically plausible.
I was under the impression that the single Californian study that purported to show that smoking prevented Parkinson's has been refuted, by a much larger Chinese study.
But I could be wrong.
As a doctor myself, I know that alot of these are THEORIES.
There is no FACTUAL link between psychosis, schitzophrenia, other mental disorders and cannabis.
In fact, there is no KNOWN cause for schitzophrenia AT ALL.
Like anything, cannabis is addictive. So is chocolate, shopping, and sex. Let's try and class those as a drug, shall we?
The reason that this storyline is so important? It all comes down to one thing - The government cannot tax cannabis.
I personally do not touch the stuff, and have no desire to.
But I have seen those who had a medical need for it benefit greatly.
I agree with the chicken/egg analagy. Its very similair to the MMR/autism debate. Autism is usually diagnosed around the same time that the MMR is given, yet we have no Factual scientific link.
I can only hope that people will stop reading poppycock stories and look for the truth!
Nope. Here's a meta-analysis here
It is a misconception that there is a distinction between theory and fact in science. There are just theories with varying levels of evidence. Correlation is rather weak evidence, to be sure, but it is evidence.
What makes a drug addictive is that it bypasses normal brain mechanisms to directly activate brain reward pathways that have evolved to support survival/reproduction behaviors such as eating, sex, and foraging (shopping). It is not true that "anything" is addictive; only certain types of drugs activate these reward pathways. Cannabis is at most weakly addictive, however.
Actually, there is probably no particular relationship between drugs referred to as "depressants" and depression. Depressants can interfere with consciousness and suppress physiological functions such as breathing, but they don't necessarily depress mood. In fact, the most reliable way to create a depressed mood seems to be to give stimulants for a while and then let them wear off.
But alcohol affects a variety of neurotransmitter systems in the brain, particularly at high doses, so it wouldn't surprise me if it produced long term changes in cognition or mood.