11
$\begingroup$

I'm not sure if questions about statistics in the context of clinical trials belong to cross validated, especially if regulatory aspects come into play.

Example for such a question would be:

Before the database lock in a (blinded) clinical trial the statistician does a blinded data review (BDR). The statistician is blinded by replacing the treatment group assignments by "random variates". One task during the blinded data review is checking distributional assumptions. And here comes the question: What does "random variates" exactly mean? Does it mean that in a trial of medication versus surgery, the term "medication" is replaced by a random number, say 7, and "surgery" by another random number, say 42, so that 7 always stands for "medication" and 42 always for "surgery"? Or does it mean that a long random series of 0's and 1's are produced and copied over the original treatment variable? The 0 could mean "medication" sometimes and sometimes it could mean "surgery" (same for 1).

The point of this distinction is: If the first replacement is done, the groups are preserved but they are only masked by a new name. If the second replacement is done, groups are not preserved. In the second case I don't see how to check e.g. a normality assumption, as the 0-group and the 1-group are potentially mixed from 2 distinct normal distributions and therefore no more look like normal distributions. So the second approach doesn't make much sense to me but it would keep the blinding more effectively because the masked groups could easily be unmasked if one could see other variables which could take on only special values in one group, e.g. the variable "complications during surgery yes/no" could unmask the 0/1-group variable.

Maybe someone even knows what's to be done because there is a clearer guideline on that somewhere...

Do you guys think that questions like this should be posted on CV? Should there be a special tag for them? Should there be a new CV-spinoff on SE (I personally find this a bad idea as it would create a too closely related ressource)?

$\endgroup$
10
$\begingroup$

Your example question seems to be one of:

  • experimental design and
  • applied statistical analysis,

both of which are listed as appropriate topics in the faq. CV is the appropriate place to ask such a question.

However, also consider weighing in on the Public Health & Epidemiology proposal on Area51. While CV is currently the best place to ask your question, I find it reasonable to argue that the concerns, methodology, and terminology of public health practice and research are sufficiently specialized to warrant a specific Q&A community (see a discussion by EpiGrad). Also, there is more to public health than the applied statistics used in the field of epidemiology. For example, questions of health law (e.g. IRB and FDA concerns for your example study) and health policy (e.g. how to use the results of your example study to affect medical practice) are not appropriate on CV nor, so far, any other SE site. Unfortunately, interest in the Public Health & Epidemiology proposal has plateaued for now.

$\endgroup$
  • $\begingroup$ thanks for your answer! I'll look at Public Health & Epidemiology and will decide if I post my question there or here at CV. $\endgroup$ – psj Jun 8 '12 at 8:28
  • $\begingroup$ @psj: Public Health & Epidemiology is just a proposal for a Q&A site, not an actual Q&A site. Post the question on CV. If you support a future Public Health site, consider participating in the proposal on Area51. $\endgroup$ – jthetzel Jun 8 '12 at 10:24
5
$\begingroup$

This is only directed towards a subset of clinical trial related questions, but if the question pertains to psychology or psychiatry, then it would be on topic at http://cogsci.stackexchange.com

$\endgroup$
4
$\begingroup$

I think that not only is it acceptable to ask questions about clinical trials and regulatory issues but also that there are many people here with the required experience to provide good answers. I for one and probably several others on CV have experience working in medical device and pharamaceutical industries and have a lot of experience working on clinical trials and facing FDA regulatory issues.

$\endgroup$
  • $\begingroup$ To whoever downvoted my answer, would you look please look at the way I revised it to see if you still think it deserves a downvote. The OP asked if it was appropriate to ask clinical trial/regulatory questions on CV. That question was answered. I was attempting to expound upon that by saying that not only is it permissible to do so but also there are enough authroities on the subject here that he could expect good asnwers. $\endgroup$ – Michael R. Chernick Jun 19 '12 at 17:46

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .