The latest PNEJM reports a large trial in adults (over 7000 patients randomized) in a very tasty trial, known as TASTE (Thrombus Aspiration in ST-Elevation Myocardial Infarction in Scandinavia).
The trial was among patients who were undergoing coronary angiography and intervention for an acute ST elevation Myocardial Infarction, whose data were already being entered into the SWEDEHEART database (I love that acronym, which stands for Swedish Web System for Enhancement and Development of Evidence-based Care in Heart Disease Evaluated According to Recommended Therapies, not bad for an acronym in a foreign language, but I guess most Swedes speak English as well as many native speakers) In one group they sucked out the clot before proceeding to stent, in the other group they did not.
Basically the study found no difference between the groups. The reason I though I would write about it is that this is an amazing study. Of the 7200-odd patients randomized the losses to follow up were zero! They managed to enroll about 2/3 of eligible patients, and because the patients were part of the registry they were able to really find out what happened to the non-enrolled patients as well. So they present results of the 4,700 non-enrolled patients as well as the randomized patients.
As this is an emergency procedure and you can’t spend a lot of time faffing about getting consent, they initially got oral consent, then randomized the patients, performed the procedure and then came back to get full signed consent later.
All of the data being collected was already being collected by the SWEDEHEARTs, so they were able to finish the study in record time for very low cost.
I think this is the way to go, we already have some well established comprehensive databases in Canada, Australia/NZ, VON, who could perform studies like this, as the accompanying editorial in the PNEJM puts it this the next disruptive technology in medical research. We could use our neonatal databases to do research which will be done faster, at lower cost, include a greater proportion of eligible babies and therefore be more generalizable. I think this would work best for comparative effectiveness research, if you are comparing things that are already being done in different centers then there is much less need for serious adverse event forms, over-detailed case report forms, and all of the structure that is imposed on clinical trials nowadays.
One other point, for those who think that being treated according to your doctor’s best guess is preferable to being in an RCT, the randomized patients had an overall mortality of 3%, the patients who were not enrolled and treated according to their doctors best guess (about sucking the clot out or not) had a mortality of over 10%. 6% of the patients who were randomized and treated according to the protocol developed heart failure, and over 10% of those treated outside the protocol. You are much better off being randomized and treated according to protocol.