study of atopic disorders in childhood before psychotic experiences in ....

[eragon]

http://www.schres-journal.com/article/S0920-9964(13)00520-3/abstract?rss=yes

[sneaker]

Thanks, eragon, for posting this study.

That is some theory they have.  Are they looking for something that does not exist and maybe have preconceived notions?  Or maybe the opposite.

Some problems they admit to is asking parents about asthma, ezcema and I think not requiring documentation.  Also, medications could be an issue.  And I think they mentioned something about outliers (?).  It was a long read, can't remember.

Also of course if people have atopic and PE, it does not mean one condition caused the other or is even connected.

Could someone explain this?

[twinturbo]

If you're worried about this being a smoking gun that asthma causes psychotic experiences--don't. In general most studies like this are correlative. They look at the instance of one and see if there is some relationship with the other in rate of occurrence. Sometimes there's good math on it, sometimes not. Sometimes there's good methods used in the study, sometimes not. Same for population, conclusions, argument.

Few look at causation, and still fewer make the effort to look at reverse causation to rule it out. Even the most perfect experiment, or findings, or what have you needs to be replicable with same or similar results when performed again. There are areas of the world where asthma approaches 10% of the current childhood population as a whole. Those 10% (or greater) are not expected to have psychotic experiences.

Markers fluctuate. They indicate a likeliness that a condition is possible. I have a marker for inflammation that may indicate rheumatoid arthritis. My doctor has the same marker with fluctuation. Neither of us have any symptom of RA. Therefore we don't chase it down.

While inflammation markers may be present it doesn't mean one caused the other or has anything to do with. If there is something that bears out in the data it will be replicated and enhanced until a strong causative is shown with reverse causation ruled out.

[CMdeux]

LOL-- well, I think you've done quite well deconstructing the problems inherent in retrospective studies of this nature, myself.

I don't (personally) think that folks in behavioral science have much business treading into medical diagnoses without having expert input, myself...    so I tend to take ALL of the studies of this type (that cancer patients suffer more depression after chemo, etc. etc.) as highly speculative and mostly a way to mine the hell out of correlation fallacies.

Take one UNCOMMON condition, and go looking for coincident common conditions, and yeah, you're probably going to find something unless your study cohort is HUGE.  Add in retrospective elements of bias, and just a smidge of possible genetic predisposition to neurosis and you'll probably hit the jackpot with something like this.

People with MS are more likely to have been diagnosed with clinical depression before they get an MS diagnosis, too.  Correlation?  Sure-- but depression probably doesn't cause MS.

The one exception that I can think of is some of the larger-scale studies that link Alzheimer's with some intriguing early warning signs, and those that link clinical depression very tightly with the development of Type II diabetes. 

[twinturbo]

At this point we're all going to have to have second degrees in discrete mathematics to wade through the fallacies.

[catelyn]

I actually find it fascinating that this is being looked at.  IMHO there is WAY more to food and the way our bodies use and react to than science realizes as you all well know. 

[twinturbo]

Most people are interested in avenues of discovery. But specious arguments aren't sufficient to constitute discovery. There's very little point to poorly performed science and there are very real problems in the mathematical models used in papers on human subjects. In other words no one really bothers to punch down into causative (or its reverse). As long as it never gets past superficial correlative there's not going to be much of the fascinating discovery we're all looking for.

[catelyn]

Depression might not cause MS, but food might just cause both depression and MS. 

I have rampant psychiatric issues in my family (overt schizophrenia).  My grandmother was branded crazy.  She was more than likely schizophrenic as are two of my cousins.  Much later I am dx with celiac.  I look back and see my family is riddled with it.  There is more than a "suspicous" argument between these two disorders.  It seems tip of the iceberg to me.

http://www.psychologytoday.com/blog/evolutionary-psychiatry/201103/wheat-and-schizophrenia-0

Have you never wondered what ELSE these allergic reactions are doing long term to your kid?  I certainly have. 

[twinturbo]

Specious, not suspicious.

I live with two published researchers. Admittedly I look at research through their lens, mostly because we read papers together.

[catelyn]

My dyslexic mistake. 

Gah maybe it would have been better to be born in 1500 and die of plague at 25 than to have all this time to think. 

[twinturbo]

Well, in truth I agree with all you're saying which is we all need to keep an open mind, a reminder we all need. My problem with all this and I do mean my problem is to go down any of these roads I have to run the gauntlet known as Dr. DH and Dr. FIL who, due to their respective fields, must publish in their discipline's journals according to established standards which do not allow the more elementary forms of modeling in correlative studies for publication. Human subject medical studies don't often get the more advanced mathematical modeling which just seems weird to me but that's as far as my personal knowledge goes on the subject. Sky and CM and others know more about statistics and populations. Thankfully.

[CMdeux]

Exactly.  Two relatively uncommon conditions, seemingly unrelated to one another mechanistically and symptomatically...

seem to more commonly exist together. 

Okay, well, there are a NUMBER of different possible explanations for that observation of correlation.  Roughly in order of probability:

1.  It's a mistake in data analysis/interpretation and when you re-evaluate, it goes away because you fixed your math error,
2.  It's pure chance and your effect goes away if you look at a different population and/or larger sample-- or if anyone else tries to see the same thing please note that THIS is about the step at which desensitization is 'stuck' in terms of investigation at the moment-- it's in replication phase,
3.  A THIRD, unobserved factor is tied to both things (and this relates to everything from here on out, btw)
4.  Both things are related to the same locus genetically-- that is, they aren't the SAME gene, but on a nearby one that also moves along with it-- as a "chunk" of related instructions-- like being fair-skinned and blue-eyed,
5.  Some underlying causative mechanism is at the bottom of both things.

The last one is the "zebra" explanation.  For it to even be worth exploring, ALL of the other items need to have been ruled out-- because any of them is more probable as explanations go. 

Most of the time it's just not that exotic.


I have a hard time with stuff like this, as well-- it's not that I don't think that causation is interesting, but this correlation parlor game just makes me NUTS.  _{Truly, how easy must it be to get tenure in this field...}

So I live with Dr. DH, but I live with Dr. ME, too-- I can ignore DH if I want to.  The angel of my own (critically thinking) better nature just won't shut up, though.  {sigh}  Takes all of the joy out of it, I tell ya.

[sneaker]

I want to keep an open mind too.

It's good to be innovative.  The authors even note they are the first, to their knowledge, to do a longitudinal study like this.

But I also agree that the study must be done right.  Not saying this one was not, I just do not know.  Because otherwise you can connect lots of things and come up with wrong results.