I’ve been surprised by how often I’m asked to run new analyses from clinical trials that include variables or groupings that were not originally reported. Aside from the time commitment involved in that, it seems problematic to include findings from new and otherwise unreported analyses.

I estimate that I receive 4-6 requests per year now related to meta-analyses that ask for some sort of reanalysis. I’m okay with simple things like “hey can you tell us what the r value is for men separate from women in this one study you did?”

I can think of many instances where the information provided in a meta analysis cannot be traced back to the original publication. How often are “ghost findings” included in meta-analyses?

A question I’ve wondered about this: if the original study can only meet inclusion criteria by running analyses that were not previously conducted or don’t flow naturally from the original study aims, is it appropriate to include in a meta-analysis?

We used the Stanley-Brown form.

CRP’s advantage held at 6 and 12 months although this advantage was not statistically significant. CRP also had around 50% fewer suicide attempts than safety planning.

Overall results support the superiority of CRP vs safety planning. This is the second RCT to support CRP’s effectiveness.

The latter. The rationale is that the big gains have already been made so the effect sizes going forward will be much smaller by default. I think it reflects a bias that views larger effect sizes as more impactful.

I should note that I’m making this point not only as someone proposing tx refinements but also as a fellow reviewer who has rebutted this perspective more than once.

The rationale presented is often that the treatment is already better than something else with a much larger effect size, so any refinements by default will have smaller effect sizes, which is viewed as having low impact.

This issue is especially pronounced in suicide prevention. I’m stunned by how often reviewers discount efforts to make an effective tx even more effective. It’s akin to saying “20% of patients attempting suicide in tx is okay. Trying to reduce that to 10% isn’t meaningful.”

Me: okay, here you go.

University: rejected. The amount charged doesn’t match our calculations of exchange rates.

Me: 🤷🏻‍♂️ that’s what I paid.

University: we’re only going to reimburse you 99.999% of your voucher total.

Me: Ok…. So you spent hundreds of dollars to save <$10. Cool cool cool.

It’s been a few years since that happened to me. I do wonder how that happens.

Did anyone by chance give you a sock?