Thursday, October 1, 2009

Maternal Smoking During Pregnancy Causes Schizophrenia?

Well, this sounds pretty far-fetched. Based on a longitudinal study in the UK, the children of women who smoked while pregnant ended up with greater likelihood of "psychotic symptoms" at the age of 12.

"Smoking in pregnancy risks psychotic children"
http://www.reuters.com/article/healthNews/idUSTRE5900XU20091001
http://bjp.rcpsych.org/cgi/reprint/195/4/294

If this relation is true, it is great to discover this clue, and start evaluating the issue more thoroughly.

But this type of study is ripe for yielding misleading conclusions. Let's take a look to see if the study data and design are worthwhile, or if someone is just out there trying to spin data for a headline.

This is based on a survey across time. The data are simply based on observing people across time and seeing what mathematically is related to what. There is no scrap of method or evidence beyond that which ties smoking and later poor outcomes in children.

I am sure any of us reading this headline can think of ways that this smoking-psychoticism relation might pop up due to some third factor.

Maybe the tendency for psychotic symptoms is partly genetic, and those moms who tend to be more psychotic tend to smoke more, and so what you are really seeing is mom smoking as a substitute indicator of the genetic likelihood of psychotic symptoms. Like pants size could be a fair substitute for body weight. Sure, the two are related, but no one would ever say maternal pants size causes obesity in her children.

These types of findings pop up in longitudinal surveys in two ways: one, there really is a relation, and two, there is a mathematical relation, but it just happened by chance. Sure, every now and then you end up meeting someone who has your same birthday. How funny! What a cooncidence. Well, it is bound to happen by chance every now and then.

Add enough measures to a longitudinal survey, and you eventually will find a relation between maternal smoking and something. Favorite color, desert preference, whatever.
Also, the authors lost a chance to help the readers, like you and me, understand what this outcome measure was: "adolescent psychotic symptoms." If the measure includes at least one symptom that could be misinterpreted, then the whoel finding falls apart. So, I pulled up the study to look at this outcome measure.

A composite outcome can yield a finding in one of two ways: one: a certain symptom carried the laod for the whole category, and two: the various symptoms all add up to a finding when they are summed.

A diet analogy: if caffeine causes indigestion, but other beverages don't, and you sum caffeine and all other beverages into a "beverage" category, then you will find that "beverages" cause indigestion.

Similarly here: are we talking about hearing voices that aren't there? Or are we talkign about some less impressive "symptom," such as "does your child ever act in nonsensical ways?" Sure a psychotic person talks in nonsensical ways, but a lot of parents of teens would answer "yes" to the nonsensical-talk question.

Turns out that on this aspect, the study is weak. First, the study does not report the actual questions.

Therefore, we cannot decently judge the quality of this study based on what is printed. In 7 pages, they could not find it within themselves to tell us the 12 "psychoticism" questions.
Next: the study does not report a separate analysis of the relation between maternal smoking and individual questions. So, like the "beverage" analogy, we cannot tell if the whole suspected relation between mom smoking and kid being "psychotic" is being carried wholely by some debatable item, such as nonsensical talk, or by some "Schneiderian, first rank" symptom of schizophrenia - such as hearing voices that are not there.

Why not do the study this way? Here, for me, with spending just a few minutes, I have asked a couple questions of this study, and sketched out a way to where the results mean nothing, yet the paper itself does not contain the easily available two pieces of info needed to determine whether we have a new reason to worry about maternal smoking, or if we just have something to put in the headlines on a slow news day.

Would it be that much trouble to add a paragraph or two, and really clear up this fatal flaw?
What are the items?

The paper describes what the items are ABOUT, but doesn't give the actual items. Are they top secret? Patented? Does this paper say somewhere, "I could tell you the questions, but then I would have to kill you?"

The paper does describe the items: "12 core questions covering halluciniations (visual and auditory); delusions (delusions of being spied on, persecution, thoughts being read, reference, control, grandiose ability and other unspecified delusions); and experiences of thought interference (thought broadcasting, insertion and withdrawal)."

Buried deep in the end of the "results" section is some evidence that my suspicion may be correct: that the relation may have emerged because of "soft," more ambiguous "psychotic symptoms."--

The authors looked at two more refined subsets of psychotic symptom questions.One subset: "definite, frequent" symptoms: those occurring more than once per month: they failed to find a relation between mom smoking and this more frequently occurring set of symtptoms.The other subset: "bizarre" items -- yes, the things like hearing voices, but not the things that, let's face it, kind of define adolescence - delusions of grandeur, delusions that various things (advertisements, songs, new stories, etc.) refer specifically to the teen, impression of being spied on, etc. --When these "softer" symptoms were excluded, and only the frankly bizarre symptoms were included, there was no relation between mom smoking and subsequent psychotic symtoms.

--Please note: this is MY guess regarding what is "bizarre" and what is "soft," partly based on my personal life experience interviewing dozens of people who actually answer a clear "yes" to these various questions, and my life experience working with dozens of adolescents with "emotional disturbance."

So, in the end, the story is good to generate a few headline clicks.

If a few pregnant women quit smoking, then that is great, too.

But I don't believe it is right to scare moms based on lousy info. There are plenty of reasons not to smoke during pregnancy without having to rely on this sketcky study.

Another, less drastic observation: I don't know what disorder these symptoms might represent, but they are not really the "prodromal" symptoms of schizophrenia -- the more subtle symptoms you might notice before the schizophrenai really shows up - which tends to happen beginning in mid-high school, or can show up first in college-age, and typically as late as later 20s. So, the symptoms detected here could be some schizophrenia, but some are surely something else. What, I don't know. But something else.

That is my conclusion. You are free to go click another blog, now.

Stick around, though, if you want my PSA.

My Public service Message: For many risks to a developing baby in pregnancy, quitting smoking, quitting alcohol, taking folic acid, etc., are behaviors that are changed TOO LATE to MAXIMALLY protect the baby if changed at the moment of the home pregnancy test.
AND half of babies are conceived to "unexpecting" moms. You young ladies out there need to behave in a healthy way long before the EPT dot goes blue. Or whatever color it goes. By time you find out you are pregnant, you have already exposed your developing baby to the risks you will now address.

Women "of child-bearing age" need to behave in a baby-healthy way all the time. Cuz, like I said, the data say that half of pregnancies are unexpected, so you can't wait til you "decide" to go get pregnant, or the EPT goes positive -- you will have missed out on some opportunity to lower risks. Look - its not me being mean. That is what the data say.

No comments: