Randomization

[DynamicDidactic]

---

Members don't see this ad.

Today I was checking my google scholar alerts and this article came through:

A Randomized Controlled Study of a Psychoeducation Group Intervention for Patients With Borderline Personality Disorder.

A Clinical Trial of a Psychoeducation Group Intervention for Patients With Borderline Personality Disorder - PubMed

This study shows that a PEG intervention can be an effective treatment for patients with BPD. The overall cost benefits of group interventions and the the applicability of a PEG intervention to underserved patients demonstrate its potential to address significant public health needs.

www.ncbi.nlm.nih.gov

As I read the abstract, I noticed this line:

"Of 96 outpatients who met DSM-IV criteria for BPD, the first 48 received the experimental treatment, whereas the next 48 were assigned to a wait list."

For the past hour, I have been questioning whether I have completely misunderstood (and in turn have taught) the term "randomized" incorrectly for all these years. Or this is another example of glaring problems with the peer-review process.

Am I taking ****oo pills?

 

[bcliff]

That seems like simple "assignment" not "random assignment." I had a conversation related to this with a colleague last year, when they indicated that they would "randomly select" cases by selecting every 5th case.

 

[deleted343839]

Controlled, yes. Randomized, not really.

The authors are Italian so it's possible there was a translation or usage issue that no one bothered to check out. But the wording of the abstract makes this seem unlikely. I have access to the journal through my library but I couldn't pull up any full text from 2020 for some reason.

 

[ClinicalABA]

...For the past hour, I have been questioning whether I have completely misunderstood (and in turn have taught) the term "randomized" incorrectly for all these years. Or this is another example of glaring problems with the peer-review process.

I don't think you can really tell just from the language in the abstract, a the group labels "first" and "next" don't convey enough information about how the groups were constructed (nor about the timing or sequence of treatments). the language is problematic though- there should be a preceding sentence along the lines of "subjects were randomly assigned to one of two groups."

Am I taking ****oo pills?

There's likely a better explanation of your symptoms

 

[DynamicDidactic]

I don't think you can really tell just from the language in the abstract, a the group labels "first" and "next" don't convey enough information about how the groups were constructed (nor about the timing or sequence of treatments). the language is problematic though- there should be a preceding sentence along the lines of "subjects were randomly assigned to one of two groups."

I read the full text. Definitely a non-random assignment. First 48 that met the criteria assigned to one goup and then the last 48 that met criteria assigned to another group.

Also, the last author is from the states and is a big BPD researcher.

 

[ClinicalABA]

I read the full text. Definitely a non-random assignment. First 48 that met the criteria assigned to one goup and then the last 48 that met criteria assigned to another group.

Also, the last author is from the states and is a big BPD researcher.

Wow! That seems like a bit of a problem, both experimentally and editorially. Certainly doesn’t meet standard definitions of “random assignment” (i.e. each participant has an equal probability of ending up in each group).

 

[AcronymAllergy]

I guess you could try arguing the semantics and philosophy behind "random" in that paradigm...as in, "they randomly presented to the clinic of their own accord when they saw fit, so why would I need to randomize them further?"

...that's all I've got.

 

[psych.meout]

I guess you could try arguing the semantics and philosophy behind "random" in that paradigm...as in, "they randomly presented to the clinic of their own accord when they saw fit, so why would I need to randomize them further?"

...that's all I've got.

You're a generous person.

 

[deleted343839]

Just a massive editorial fail, really.

 

[psych.meout]

Just a massive editorial fail, really.

Yeah, it's not even nitpicking some minor detail buried in the text, it's the damned title.

 

[DynamicDidactic]

Anyone want to email the editor?

 

[Ollie123]

Definitely not randomization. Sometimes reasonable to do if for whatever reason you cannot do true randomization....but you don't call it randomization.

Either reviewers or editors screwed up on this one though.

 

[Fan_of_Meehl]

I guess you could try arguing the semantics and philosophy behind "random" in that paradigm...as in, "they randomly presented to the clinic of their own accord when they saw fit, so why would I need to randomize them further?"

...that's all I've got.

LOL.

 

[ClinicalABA]

Not my population or type of work, but for those of you familiar with outpatient work and BPD, what do you think are some potential factors that might lead to differential responses to treatment between the first 48 to show up versus the second 48?

 

[cara susanna]

Not my population or type of work, but for those of you familiar with outpatient work and BPD, what do you think are some potential factors that might lead to differential responses to treatment between the first 48 to show up versus the second 48?

Um... better time management skills? More motivation? Perhaps those who arrived earlier had fewer obligations that could interfere with treatment engagement, like work or childcare?

 

[Fan_of_Meehl]

Also may have had different staff evaluating/ screening them. Innumerable potential confounds that we may not even be able to speculate about or predict.

 

[AcronymAllergy]

Not my population or type of work, but for those of you familiar with outpatient work and BPD, what do you think are some potential factors that might lead to differential responses to treatment between the first 48 to show up versus the second 48?

I admittedly didn't look at the paper, so this might be a moot point, but also possible cohort effects, depending on how long a period of time the first vs. second 48 took to come through. Time of year, recent area-wide stressors, changes or trends in possible referral sources, etc.

 

[DynamicDidactic]

Time of year

Innumerable potential confounds that we may not even be able to speculate about or predict.

Yup.