A new publication about SUPPORT? Haven’t we heard enough?
Well no, this is fascinating, although not entirely unexpected if you think about it.
One of the 2 comparisons of the SUPPORT trial involved randomization to be either immediately intubated to receive surfactant, or to attempt to keep the baby extubated and on CPAP, unless they exceeded 50% oxygen, or just weren’t able to breathe. At the time, some centers were intubating routinely, based on older trials which showed benefit, but which did not use what we now consider best treatment in their comparison groups. Some centers were trying to keep babies on CPAP, based on prior data from the COIN trial, and on accumulating experience elsewhere. In other words this also was a comparative effectiveness comparison, both approaches were considered within acceptable medical care approaches and both were in common use.
Some centers in SUPPORT had less experience with avoidance of intubation, and starting early CPAP in the DR, so when SUPPORT was active they started to gain more experience with that approach.
This new article shows that babies who were not actually in the trial, but were born in a participating hospital while SUPPORT was being carried out, ended up being treated differently than before the trial.
LeVan JM, Wyckoff MH, Ahn C, Heyne R, Sánchez PJ, Chalak L, Jaleel MA, Burchfield PJ, Christie L, Soll R et al: Change in care among nonenrolled patients during and after a randomized trial. Pediatrics 2013.
Babies born at the Parkland hospital who were under 28 weeks were intubated in the DR 85% of the time before SUPPORT, babies who were not in the trial were intubated 61% of the time while the trial was on-going, and after SUPPORT had ended the frequency stayed at 61%. This wasn’t a change that everyone adopted at the same time, babies in the Vermont Oxford Network, not involved in SUPPORT, over the same interval did not have much change in their intubation frequency.
This ‘spill-over’ effect is not unexpected, (it is one of the reasons that we sometimes do cluster randomized trials) but this is a very clear example.
One implication of this is that we will need to have an informed consent process for all the non-enrolled patients in clinical trials. As has been pointed out, consent for research and for clinical care have the same requirements, so we will need to inform parents that there are trials gong on in our NICU at the present time, which may change the care that their babies receive whether or not they are enrolled in the trial. Maybe also the consent forms for the trials should have a section which notes that, even if the parents refuse the trial, the performance of the trial might change how their baby gets treated anyway… so they might as well sign up to the trial! How could this happen? Surely doctors know what is optimal care, how could the fiduciary obligation of a doctor to provide optimal care to his patients, who are not in a trial, be affected by a trial being performed? (Paragraph written with tongue firmly planted in cheek: if you are not a native english speaker that means that I am not really serious).
This is still more evidence that we are already in a learning health care system, approaches to care change as a result of many factors other than published best evidence, when you have more exposure and more experience with a particular approach you may start to use it more, even if the evidence does not (or not yet) show substantial advantages. Maybe this is appropriate, maybe not, but it is a reality of modern health care.