@ProfDFrancis
Top tip for fellows:
If you are unable to freely access a PDF of a study say, conducted using public money, written by freely by researchers, and peer-reviewed for free by others…
Under no circumstance copy the Pubmed ID and then paste into
sci-hub. tw
to easily download
🚨
#ACC24
Breaking News 🚨
The ACC.24 Late-Breaking Clinical Trials are out! Some of the trials being presented are:
◼️ RELIEVE-HF
◼️ EMPACT-MI
◼️ AEGIS-II
◼️ VICTORION-INITIATE
◼️ Step-HFpEF Dm
◼️ TELE-ACS
Learn more & see the full list of trials:
Presented at
#AHA23
:
ORBITA-2 trial: In patients with stable angina who were receiving little or no antianginal medication and had documented ischemia, PCI resulted in a better health status with respect to angina than placebo at 12 weeks. Full results:
@ProfDFrancis
@drjohnm
I can trump your video - with this video from when the UK rolled out thrombolysis for acute stroke. (Watch with sound). I used it in a presentation in 2006. Anyway - the data for thrombectomy looks good - I think the UK is working out how/who to provide the service
@LCC_DrMalik
@venkmurthy
@f2harrell
@NEJM
1) Here is an R-notebook with the data and the “rms” style 7-level ordinal analysis. Hope I got it right!
2) You can customise the analysis to fit your personal ordering of the mRS (i.e. you may consider some states worse than death).
Interesting editorial (and paper in circ).
Top tip for fellows. Don’t google the first two references in it for the back story. It’s not at all interesting.
Though if Santa would like to bring us funding for an RCT as well that would be great... we have all been very good this year and almost finished recruitment for HOPE-HF
#dontdisthehis
@f2harrell
@NEJM
Hypothetical numbers: No treatment - 50% have events; Rx A -10%, Rx B - 5%. A is tablet, needs monitoring; B is injection, painful. If C has event rate of 7.5% (or even 10%); but oral and no monitoring patient may prefer. The unknown anchoring rate of no treatment may be key.
@venkmurthy
@f2harrell
@NEJM
The raw data for the outcome would be a table of TrialID, the treatment arm, and the mRS at 30 days. This can be constructed from the figure and analysed. Click my notebook link -> took me 5 mins
P.s Amazing Trial.
@statsepi
@ADAlthousePhD
Nice line snuck in at the end - “In summary, authors should not attempt to get cute when performing survival analysis using composite endpoints. “
📌 This Friday, Dec 18, Dr.
@mshunshin
of
@ImperialNHLI
will present findings from
#SAMSONtrial
, an n-of-1 trial to assess side effects of
#statins
compared with placebo and no treatment. Join us for our final Grand Rounds of the year!
🔗
#pctGR
@barttels2
@anupampom
@f2harrell
@venkmurthy
@NEJM
In this case I think it is for the intellectual exercise. More interesting to learn with actual data, and practice techniques + methods so that you can pre-specify them in your own work before you see the data. Message is 1) Amazing work by authors 2) Don’t have an OOH arrest.
@f2harrell
@GreggWStone
Ask patients the relative (dis)utility of each end-point and re-weight existing RCTs of coronary intervention using those results. is one attempt (by us).
@omaclaren
@stephensenn
If you want to have a neurosurgery hobby, start with this paper - Emergency burr holes “how to do it”. It features the invaluable advice "It may be useful to nominate a person to read out the components of this guide as the procedure is being performed”
@GreggWStone
@f2harrell
If this were true for all hypersensitivity reactions, then desensitisation protocols - e.g. for Asprin pre-PCI; or test doses of amphotericin - wouldn't work. Thanks to
@JKAronson
(who taught me this at medschool) -
@AdeMarvao
@DrAfzalSohaib
Mechanism, haemodynamics, and physiology make sense - but need RCT. Our HOPE-HF trial for HF with long PR - but need more RCT for other Dz
@EJSMD
@SchakrabartiEP
@SergioPinski
@MDT_Cardiac
To be fair to the trialists and
@MDT_Cardiac
it was a fantastic idea for a trial, but they estimated they would need to randomise 2300 subjects and only got 44 patients in 13 months. It showed willingness to fund and try - but sometimes you have to stop and re-consider.
@BSHeartFailure
Here is a clickable link - open access as well - so no need to sign in.
Uses non-invasive, beat-by-beat finger BP to optimise CRT. Now built into the
@finapres
machine - so easy-peasy.
Using this technique in the HOPE-HF trial
#dontdisthehis
@ncurzen
If you have lot (9 segments!) of ischaemia and you don’t get sent for revasc it likely means there is something else very wrong with you - hence higher mortality. If you force the lines to be linear - then they will cross.
@DavidLBrownMD
@wpentz
@drjohnm
The audience might like to read the authors’ reply to the Letters to the Editor. Both educational and entertaining, and deals with some of these “issues"
"Applying a positive spin could have smoothed the reception of the trial, but as authors we have a duty to preserve scientific integrity."
#ORBITA
authors respond to critical letters in
@TheLancet
@ProfDFrancis
@DrVikasSaini
@DanMarkMD
@ajaykirtane
Change seen in the placebo group is not the placebo effect, just as the change seen in the intervention group is not the intervention effect
@ADAlthousePhD
@boback
@JeremySussman
STS and similar typically will be calculated away from the patient - e.g. the fellow prepping the MDT form in the evening for the following day. CHADS2-VASc calculated in head whilst taking history in clinic with patient with new AF. Hence the difference in complexity.
@statsepi
the give away is that they talk about selecting the right patient for the procedure rather than the right procedure for the patient after a non-positive RCT.
The SDD (how much it may vary from one reading to another) for SBP on a 24h holter is around 11mmhg (e.g. from placebo arm of spyral off-med). Titrating to a 10 mmHg range will lead to chasing your tail.
#ESH
@dr_benoy_n_shah
@venkmurthy
@vass_vassiliou
@MarcDweck
Re: RCTs - to really show that the marker should guide therapy you want to recruit all-comers (those with high and low marker values), randomise them to the therapy, then at analysis show there is an interaction between the biomarker at baseline and treatment effect. 1/n
@scleroplex
@ProfDFrancis
Exactly, when people e-mail me for a paper, because I 100% have them nicely organised on my computer, I can send it to them immediately.
I don’t, for example, have to then find my own paper in Pubmed, copy it into sci-hub, and then send it to the requester.
@GwilliamsPh
@PH_Evidence
We are looking forward to submitting a full paper soon - with a lot of more analysis of the individual patient results. We have some great graphs and visualisations ready!
@DrFuSiongNg
@DrAfzalSohaib
Did patients at some centres really have to pay for assigned treatment? Primary source? If so that is all the more reason you cannot trust the ‘as treated’ analysis as the richer you are the better your outcomes are irrespective of treatment.
#CABANA
@drjohnm
@hswapnil
Re: ICD with NICM. It is likely that ICDs now are just (or more so) efficacious at terminating spontaneous otherwise fatal tachyarrhythmias. The RR/HR/OR (forgive me
@f2harrell
) for this specific death should be ~preserved. The baseline incidence will have decreased with time...