RCTs Can Inform Even the Most Dire Medical Situations
I've heard people say you can't randomize extremely ill patients. A trial presented at ESC proves this idea false
Investigators from Leipzig Germany called the trial ECLS-SHOCK. It studied the use of extracorporeal life support in patients with cardiogenic shock due to acute myocardial infarction.
ECLS-SHOCK delivered shocking results. First some background.
Cardiogenic shock is medical jargon to describe the really bad situation when the heart cannot deliver enough blood to the body. Think pump failure. It’s bad because within minutes to hours, your organs start to die from lack of oxygen and nutrients. Cardiogenic shock often leads to death.
A common cause of pump failure is acute myocardial infarction—or heart attack. A blocked coronary artery causes the downstream muscle to stop squeezing. If the amount of damaged muscle is big enough, the pump fails.
My colleagues who do interventions can open these blocked coronaries, which can restore pump function, but sometimes it takes time for the heart to recover. During that time, cardiogenic shock can kill the patient.
This led to the idea of mechanical circulatory support—which is a way to support the failing heart while muscle is recovering.
The first attempt at this sort of support is with an intra-aortic balloon pump. I spent many hours during training learning the ins and outs of the IABP.
Then in 2012, the same authors of ECLS-SHOCK, presented results of the IABP-SHOCK II Trial. IABP vs standard care in patients with cardiogenic shock. Answer: no difference. The IABP supports the heart. It makes nice noises and improves the blood pressure tracings, but 30 day survival was the same.
ECLS-SHOCK studied what many thought was a much better support tool.
ECLS is also called veno-arterial ecmo or extracorporeal membrane oxygenation. I will oversimplify it, but, basically, this takes blood out of the body, runs it through an oxygenator machine and pumps it back to the body. Think—mini heart-lung machine.
ECLS is invasive. It requires large bore access to central veins and the femoral artery. That’s a problem because these patients just had stents placed and have had high dose antiplatelet drugs. Think bleeding. ECLS also requires a highly trained team to adjust the flows and monitor the patient.
ECLS is quite popular. It’s use has risen 10x over the past 10 years.
But. But. We had no evidence that it reduced death compared to standard care. In fact, a smaller trial, called, ECMO-CS found no difference in outcomes.
Investigators from Leipzig recruited a number of centers in Germany and Slovenia to enroll patients with cardiogenic shock due to acute MI. They randomized one group to get ECLS and the other group got standard of care.
We should pause there for a moment. This is brave. Brave because cardiogenic shock very often leads to death. Mechanical support, such as ECLS, supports the heart and some doctors would consider it unethical NOT to use it. (But regular readers of Sensible Medicine know that things that make sense or should work often don’t.)
When the ECLS-SHOCK authors counted up deaths at 30 days (the primary endpoint), they found that 47.8% had died in the ECLS group vs 49% in the control arm. This is essentially the same. The hazard ratio calculated to 0.98 with confidence intervals ranging from 0.81 to 1.19. Here are the survival curves.
But there is more.
Not only did ECLS not reduce death, it also led to harm. Patients in the ECLS arm had a 2.4x greater rate of bleeding and a 2.8x greater rate of vascular complications.
Sensible Medicine is a reader-supported publication. To receive new posts and support our work, consider becoming a free or paid subscriber.
Two Big Lessons
I like to highlight studies that teach multiple lessons. Here there are specific and general lessons.
The specific take-home is that ECLS does not work in cardiogenic shock due to acute MI. Going into this trial, a previous study had found no difference. We had pessimistic priors. Now, a much larger trial, also finds no survival advantage and much higher rates of complications. This is really important information for cardiologists.
The general lessons, though, are much more valuable. The ECLS-SHOCK authors have reminded us in medicine to always be cautious of doing things that make sense—even in dire situations.
Yes, on paper, ECLS, (like the IABP), supports a failing pump. It’s a totally plausible intervention. But plausibility is too low a bar.
Treating people is not like fixing machines. We are complicated. Treatments often lead to unintended consequences.
In medicine, you have to show something works to reduce an important outcome in real patients.
If Professor Thiele and his colleagues had not been curious and brave, we would be doing this invasive costly procedure for years. And. We would be harming patients and not knowing it.
ECLS-SHOCK is a shining example of why we need a culture of randomization in this profession. If you can randomize patients dying of pump failure, you can surely randomize many if not most therapy choices.
Sensible Medicine remains an advertisement-free source of medical information. We appreciate your generous support. Thank you. Feel free to add comments. JMM