The transcript of yesterday’s discussion is now available in PDF format here. If you wish to download the PDF, just click on the green ‘Download’ button that appears at the bottom right. A few people have confirmed that the link works, but if you do have a problem, let me know, either in the comments section here or on Twitter.
I have included all tweets up until 10pm, although the conversation has been continuing since (albeit at a rather slower pace), so do look at the #TwitJC tag.
We will also be posting a summary soon, highlighting the key points that came out of the discussion, and this should be available in the next day or so.
Thanks again to everyone for making yesterday such a success!
Congratulations on getting this going!
I would think it important to emphasize going over systematic reviews more often than “landmark” papers. Such papers often were the first published in their field, often in top journals. There is ample literature that shows that, for a number of reasons, those papers often overestimate the true effect of interventions. Also, given the relationship that pharma and devices have with journals, it is important to note issues related to publication bias. Of course, this can only be assessed when looking at the overall literature, not the single paper. Finally, given that there are several ways in which we get the wrong idea, appraisal of methods of a single paper will only get you to ascertain bias, but not spin, pub bias, inconsistency across studies, etc, all of which reduce the strength of inference you can bring back to the bedside.
The solutions:
1. Start from a real question in practice – do not let the post office or the journal editorial board tell you what to read.
2. Choose the most aggregate form of the evidence to ascertain both bias as well as the other factors described above. Note that systematic reviews are themselves open to the same issues.
3. When you formulate your question decide which outcomes you are truly interested in. If not on the paper you are reviewing consider the possibility of outcomes reporting bias. Do not just accept whatever outcome is reported.
4. Beware spin: composite endpoints, surrogate markers, subgroup analyses, inadequate comparators (too much or to little of an effective alternative, or placebo when an effective alternative exists). Above all avoid the intro/discussion sections which is where most of the interpretational spin is introduced.
and above all…
5. HAVE FUN!
Well done, missed the discussion but will be joining in next week. Twitter & WordPress offer great potential for serious and educational discussion. Perhaps ask along an expert in the field to help guide the discussion, answer technical questions etc? Real life journal clubs benefit from having someone to facilitate and ensure “facts” are correct to allow interpretation….just a thought.
Awesome job guys. I’ve been thinking of an online journal club for a few weeks now and really glad you’ve taken the initiative and built this. I wonder of the scope of expanding beyond the medical field though? I’ll be watching and joining in to see how it goes!
I posted here http://is.gd/R1h649 about a slower-paced, wider ranging JC, but you’re doing an ace job and I don’t necessarily want to start another, especially if this works well!
[...] Comments « Transcript: First #TwitJC Twitter Journal Club – 5th June 2011 [...]
[...] Just a reminder that the transcript from Sunday evening is available here, should you wish to look through it. Also, please do comment on this post and continue the [...]
[...] stats and the quality of the study’s design and data. If you want, you can download the entire transcript. (And I think someone could make a killing if they could work out a way of threading such hashtag [...]