Changing the design of a research project...after IRB submission?

[Jumb0]

So, I submitted the IRB proposal for my summer research project, which I declared as a retrospective cohort study. Since it involves only chart review, we applied for full waiver of the consent process.

However, after some consideration, it appears that the cohort design may be a fruitless endeavor since we might not have a large enough sample size to reach statistical significance. I'm thinking that I may need to change the design to a case control study instead.

Will this be a huge issue?

The change in design will NOT necessitate any additional data. In fact, the IRB proposal we submitted requested more data than we will even probably end up using...Also, since we are going for total waiver of consent, the change would not affect patients in any way whatsoever...

For the record, we have NOT received approval yet. We have simply submitted our proposal and are waiting...

---

Members don't see this ad.

 

[.hematoma.]

I'd wait for approval and then submit an amendment to the IRB, in my opinion that's the easiest way to tackle this

 

[Jumb0]

I'd wait for approval and then submit an amendment to the IRB, in my opinion that's the easiest way to tackle this

Thanks for the reply. That's what I was thinking too.
What if they don't approve the current submission outright and instead ask for some additional information? Would that be a good time to let them know?

 

[.hematoma.]

Thanks for the reply. That's what I was thinking too.
What if they don't approve the current submission outright and instead ask for some additional information? Would that be a good time to let them know?

Hmm I think if you let them know they'd want to re-review it, but I'm not entirely sure.

 

[ortnakas]

Thanks for the reply. That's what I was thinking too.
What if they don't approve the current submission outright and instead ask for some additional information? Would that be a good time to let them know?

Probably, since you'll be submitting an amended application anyway. Might as well make all the changes at one (their changes and yours) in the name of efficiency.

Don't be disheartened if they do ask for additional info, by the way. I worked in research for a few years and it was very rare anything got approved 100% on the first go.

 

[mvenus929]

Thanks for the reply. That's what I was thinking too.
What if they don't approve the current submission outright and instead ask for some additional information? Would that be a good time to let them know?

The IRB sent me some questions they had about my proposal, and I had to send edits back. I tried to make a few subtle changes in the protocol (expanding the age range, since I had changed the age range on my proposal, but not on my application), and they said I couldn't submit those changes because the protocol had already been reviewed under the current settings, even though I already had to make changes to suit their questions.

So you'll likely have to submit an amendment either way.

 

[dempty]

However, after some consideration, it appears that the cohort design may be a fruitless endeavor since we might not have a large enough sample size to reach statistical significance. I'm thinking that I may need to change the design to a case control study instead.

I wanted to comment on this part briefly. Keep in mind that getting "statistical significance" shouldn't be the goal of a study. Instead, focus that effort towards obtaining a good and precise estimate of the effect size(s). If significance happens to be reached at a predetermined threshold, then it happens, but significance in itself shouldn't be a driver for decisions-- despite what a non-statistician PI might suggest (not sure if this is your case, but I've heard of it). It's part of the research problems and publication bias we see today; the odd phenomenon of only valuing studies with significant results causes us to focus on the wrong aspects.

After all, all else the same, reaching significance can be toyed with by less-than-honest or ill-informed researchers. The p-value is a function of the sample size, among other things, so practically every test can be significant with a large enough sample.