Schizophrenia: No Smoking Gun

[Note: despite how some people are spinning this, tobacco is still really really bad and you should not smoke it]

I.

Schizophrenics smoke. A lot. Depending on the study, about 60-80% of schizophrenics smoke, compared to only about 20% of the general population. And they spend on average about 27% (!) of their income on cigarettes. Even allowing that schizophrenics don’t make much income, that’s a lot of money. Sure, schizophrenics are often poor and undereducated and have other risk factors for smoking – but even after you control for this, the effect is still pretty strong.

Various people have come up with various explanations. Cognitively-minded people say that schizophrenics smoke as a maladaptive coping strategy for the anxiety caused by their condition. Pharmacologically-minded people say that schizophrenics smoke because smoking accelerates the metabolism of antipsychotic drugs and so makes their side effects go away faster. Pragmatically-minded people say that schizophrenics smoke because they’re stuck in institutions with nothing to do all day. No points for guessing what the Freudians say.

But all these theories have problems. Sure, schizophrenics are often institutionalized, but even the ones at home smoke a lot. Sure, some schizophrenics are often on antipsychotics, but even the ones who aren’t on meds smoke a lot. Sure, schizophrenics are anxious, but we don’t see people with Generalized Anxiety Disorder having 80% smoking rates.

As usual, I’m more biologically-minded, so I find it interesting that some of the genes that most commonly turn up as linked to schizophrenia – especially CHRNA3, CHRNA5, and CHRNA7 – are in nicotine receptors. Indeed, some of them are also the genes identified as risk factors for smoking. Further, there’s a lot of evidence that schizophrenic people actually feel better and have fewer symptoms when they’re smoking. Further, schizophrenics tend to gravitate toward cigarettes with higher nicotine content, and smoke them in ways that maximize nicotine absorption.

It seems like part of the problem with schizophrenia is that the brain’s nicotine system isn’t working well. Smoking supplements nicotine and makes the system run smoother, so schizophrenics feel better when they smoke and continue to do so. This is the widely accepted self-medication hypothesis.

I like this because it’s a really elegant example of…I don’t know what you’d call it…memetic evolution? Nobody knew that nicotine helped schizophrenia, nobody told the schizophrenics that, but they sort of naturally gravitated to an effective treatment for their condition by going in the direction of things that make them feel better, even going so far as to unknowingly gravitate toward cigarette brands with more nicotine. They did all of this before psychiatry had any idea why they were doing it, and in the face of constant protests that it was stupid and useless. This should be a warning to anyone who’s too quick to tell patients that their coping strategies are maladaptive.

But there’s a much more important question here: does smoking cause schizophrenia? How about prevent it?

II.

First, the causation argument. Gurillo et al do a meta-analysis and conclude that “daily tobacco use is associated with increased risk of psychosis and an earlier age of onset of psychotic illness. The possibility of a causal link between tobacco use and psychosis merits further examination”. That is, schizophrenics are already smoking much more at the moment their schizophrenia starts. This suggests that maybe smoking is helping to cause the schizophrenia?

All nice and well, except for a few things. First, this study ignores the possibility that the genes that cause schizophrenia might also cause increased smoking, even though we have some evidence that this is true (actually, it doesn’t ignore this, it mentions it, but uses it as a reason why a schizophrenia-smoking link is more plausible). Second, we know that people who will later develop schizophrenia are seen as kind of odd even before they come down with the disease, and it’s possible that they’re already in some unusual brain state that smoking helps relieve. Third, this study is not controlled – meaning that we’re totally helpless before factors like “people destined to later develop schizophrenia are often poor, and poor people smoke more”.

And fourth, another study shows exactly the opposite.

Zammit et al (thanks to @allfeelsallthetime for the tip) looks at 50,000 teenage Swedish conscripts, then follows them throughout their lives to see which ones do or don’t get schizophrenia. They find that without adjusting for confounders, smokers are more likely to get schizophrenia. But when you do adjust for confounders, smokers are less likely to get schizophrenia, (hazard ratio 0.8, p = 0.003) and heavy smokers are much less likely to get schizophrenia (hazard ratio 0.5)! A dose-dependent relationship was found between smoking and protection from schizophrenia. This is really interesting.

Why do we find such different results from these two studies? The only explanation I can think of is that the second study controls for various factors including cannabis use, personality variables, IQ, past psychiatric diagnoses, and place of upbringing (thanks @su3su2u1 for the tip) and the first study controls for zilch. In fact, we find that the second study’s uncontrolled numbers are not that different from the first study’s uncontrolled numbers, and that the only difference is that the second study then went on to control for confounders and get the opposite result. Controlling for more things is not always better, but controlling for a few things that previous studies and common sense suggest are very relevant is pretty superior to just leaving the data entirely unprocessed. Advantage very much second study.

III.

Unlike certain people on Facebook, I fucking hate science. Let me explain why.

The first study here, Gurillo et al, was published ten years after the second study. Since it is a meta-analysis, it included the second study in it. The authors of the first study definitely read the second study. They just didn’t care.

Nowhere in the first study does it say “By the way, we read this other study that got the opposite results from us, let’s try to figure out why, oh, it was because they controlled for things and we didn’t, maybe that should call our findings into question.” You know what they did do? They listed the second study as finding that smoking increased schizophrenia risk, because the rules of their meta-analysis said they would only take uncontrolled data, and so they did. You can read this entire study, which cites the second study no fewer than six times, without hearing at all about the fact that the second study got the opposite result using likely better methodology.

Then they go on to conclude that:

Cigarette smoking might be a hitherto neglected modifiable risk factor for psychosis, but confounding and reverse causality are possible. Notwithstanding, in view of the clear benefits of smoking cessation programs in this population, every effort should be made to implement change in smoking habits in this group of patients.

Clear benefits! Every effort! Aaaaaaah!

I mean, I know where they (and the Lancet editors, who write a glowing comment backing them up) are coming from. Smoking is bad because lung cancer, COPD, etc. But now we have these things called e-cigarettes! They deliver nicotine without tobacco! As far as anyone knows they carry vastly less risk of cancer, COPD, etc. If nicotine actually prevents schizophrenia rather than causing it, that is the sort of thing we should really want to know. And instead we’re just getting this “We should make schizophrenia patients stop smoking, because smoking is bad”.

Look. I am not going to come out and say that there’s great evidence that nicotine decreases schizophrenia risk. There’s one study, which other studies contradict. I happen to think that the one study looks better than its competitors, but that’s my opinion and I have nowhere near the evidence I would need to feel really strongly about this. But I feel like we are very far from the point where we know enough to be pushing people at risk of schizophrenia away from nicotine, and light-years away from the point where we can use phrases like “clear benefits”.

Possibly I am an idiot and missing something very important. But if this is true, I wish the authors of the new study, and the editors of The Lancet, would have acknowledged the existence of the conflicting study and patiently explained to their readership, many of whom are idiots like myself, “Here’s a study that looks better than ours that seems to contradict our results, but here’s why our study is nevertheless far more believable.” That’s all I ask.

No matter how much of an idiot I am, I can’t possibly imagine how that wouldn’t be a straight-out gain.

PS: Cigarette smoking definitely decreases your risk of Parkinson’s Disease. Parkinson’s is similar to schizophrenia in that both involve dopamine. But schizophrenia involves too much dopamine and Parkinson’s too little, so the analogy could go either direction.

PPS: Tobacco smoking is definitely still bad! Nothing in here at all suggests that tobacco smoking has the slightest chance of not being a terrible decision!