Prediction of Relapse in Schizophrenia paper, from Arch Gen Psych. Having read it, what they were trying to do makes more sense to me, but it's still problematic.
The paper's introduction comments:
...A major deficiency in the clinical management of schizophrenia is the lack of proven methods for predicting the subsequent course of illness and identifying those patients who require neuroleptic maintenance to prevent relapse or, alternatively, those who would remain stable for substantial periods of time without drug treatment or while receiving substantially reduced dosages.
There were some previous studies suggesting that behavioral response to psychostimulants could be used as a predictor of relapse. They chose ritalin because it preferentially inhibits uptake of dopamine, and they were trying to find support for the dopamine hypothesis in schizophrenia.
The reason given for conducting the study is:
The application of a psychostimulant provocative test using methylphenidate with stable schizophrenic outpatients undergoing neuroleptic maintenance treatment is an attempt to study whether it may be feasible to exploit this phenomenon for clinical purposes.
That's it. Vague, but it kind of makes sense. It's like a glucose challenge test. You want to know if a specific patient will relapse if you take him off his meds. So you do the ritalin challenge test, and if he responds in a certain way, you know he needs to stay on his meds. And if he responds another way, you can take him off. And what are those "ways"? That's what the study is trying to find out.
They gave most of* the subjects a ritalin infusion, and a placebo infusion, 1 week apart, in random order, while they were still taking meds. They did this in a double blind fashion. They rated the subjects before and after infusions for behavioral, physiological, and physical states and changes.
Then they repeated this process several weeks after discontinuing meds.
This is what the ratings found:
The white bars are before infusion, and the black bars are after.
You'll notice that while it was double blind, you can pretty much tell when the patient got ritalin.
Then they followed the patients for a year, or until they relapsed and needed meds again. And they looked at how response to ritalin infusion correlated with time to relapse. This is what they found:
Patients with greater behavioral and TD changes in response to ritalin, as well as patients with greater baseline TD, had significantly shorter times to relapse. Blink rate and pulse rate were not significantly correlated.
The paper does a bunch of additional analyses, with groups of symptoms rather than individual ones, to get more of what they wanted. But to me, the most interesting result is that patients with more baseline TD relapsed more quickly. That could be a useful piece of data, and it doesn't require a ritalin infusion.
There are technical problems with the study. It had very few subjects, there was a protocol change after the study had begun* (that's why some subjects only had infusions after discontinuing meds), and the paper was written before follow-up had been completed on all the subjects. Also, I can't quite figure out if they did an intent to treat analysis.
They also fudged a bit in the comments section:
...Specific biologic and clinical variables, including behavioral response to methylphenidate, presence of TD, blink-rate response to methylphenidate, and pulse rate response to methylphenidate, under specific pharmacologie conditions, are associated with outcome in terms of time to relapse following neuroleptic withdrawal.
In the results, blink rate and pulse rate are not so associated.
Here's the ethical problem. Participation in the study could not possibly have benefitted any of the subjects. It only had the potential to do harm. It's not like the authors were positing that a ritalin infusion might delay or prevent relapse.
So I'm skeptical about the consent process. We're not talking about healthy grad students who volunteer to stay awake for 3 days in exchange for $10 and meals. Did the subjects really understand that participation was completely altruistic? That the results could only benefit others, down the road, and maybe not even that?
I would have felt better about it if the idea had been addressed. They mention in the methods section that patients were evaluated for ability to give consent. But I think that somewhere, the authors should have written about the fact that the study was designed in a way that would not benefit the subjects. And that a shorter time to relapse after ritalin infusion might imply a different subgroup, one that might have had a longer time to relapse if they hadn't been given ritalin.
What do you think?