When covid-19 patients get so sick that ventilators can no longer keep them alive, doctors have one last-ditch “Hail Mary” option. It’s called extracorporeal membrane oxygenation, or ECMO, and it’s a form of lung, and sometimes heart, bypass that my colleagues and I are increasingly turning to during the pandemic.

To put a patient on ECMO bypass, we surgeons insert large tubes into a patient’s blood vessels, either in the neck, groin or directly into the heart. That gives us the ability to remove venous blood from the patient, run it through an artificial lung and then push oxygen-rich blood back into the body.

I first performed an ECMO years ago, during my surgical residency. I was called to the bedside of a patient on a ventilator who couldn’t keep his oxygen levels up. Despite the ventilator forcing oxygen-rich air into his body, his lungs were too damaged to absorb that oxygen into the bloodstream. His lungs needed time to rest and heal. He could no longer survive even on the ventilator. We decided to try ECMO. Knife in hand, I made an incision in his neck. I found the two major blood vessels, the carotid artery and jugular vein, and inserted a large tube into each. We hooked the patient up to the ECMO machine, and the color of his blood went from dark to bright red as the artificial lung infused it with oxygen. His oxygen levels increased to 100 percent. I thought to myself, why don’t we do this for every patient struggling to live on a ventilator?

But as I learned over time and after many ECMOs, the procedure is a resource-intensive, costly and risky treatment with many complications. And it may only improve survival for a small number of covid-19 patients — though the data is limited. So this raises the question: Should we be adopting widespread use of ECMO for covid-19 patients when our health-care system is struggling with a lack of simpler resources?

Right now, some hospitals across the country are discussing rationing ventilators and considering “do not resuscitate” orders for covid-19 patients because treating some of them may be futile and because we can’t protect health-care workers. On the other hand, many patients and families understandably want to try everything possible, including ECMO, even if the chances of survival are slim.

Just last week, ECMO was used to save a covid-19 patient’s life in Chicago. There are a handful of reports of ECMO being used for covid-19 patients across the country as well. Some projections have suggested that as many as 12,000 to 32,000 covid-19 patients across the United States may need ECMO, depending on how quickly the virus spreads. That would far exceed the number of ECMO for respiratory illnesses that we do yearly in this country. In response to this potential need, the U.S. Food and Drug Administration recently issued guidance to help expand the availability of devices that could be used for ECMO.

But the procedure isn’t an option for everyone. The high-risk complications, such as uncontrollable bleeding, stroke and infection — coupled with the low chance of survival even with bypass — mean that the risk often outweighs the benefit. Many of the patients that we put on ECMO suffer from acute respiratory distress syndrome (ARDS), the same condition that is making covid-19 patients sick. In non-covid patients with ARDS, studies show that survival with ECMO is about 50 percent in carefully selected patients, and a number of patients who survive suffer disabling complications. Whether these numbers apply to covid-19 patients is unclear.

Initial data from Wuhan, China, suggest a very poor outcome for covid-19 patients put on ECMO. In that study, 11.5 percent of covid-19 patients needed ECMO and only one of the six survived. In another study out of Wuhan, 2 percent of patients (three people in total) needed ECMO, but none survived. Numbers from the extracorporeal life support organization show that 275 covid-19 patients worldwide have received ECMO so far, the majority of whom are still on ECMO. Only 28 percent of them have been discharged from the hospital. These numbers are incomplete estimates, however; as only a subset of centers participate in this registry. Despite the limited data, the World Health Organization has recommended that consideration be given to using ECMO for covid-19 patients if resources permit.

But it’s one of the costliest and most resource-intensive treatments in the hospital, with daily need of blood products, medications, extra health-care workers to run the machine and PPE for workers. Many patients remain on the bypass machine for days to weeks. This translates to many orders of magnitude more resources than a ventilator and significant utilization of health-care workers.

ECMO isn’t available at every hospital, either. The latest figures show that 264 U.S. hospitals out of more than 6,000 in the country have the ability to perform ECMO. Because of geographic distribution, only 58.5 percent of the adult population in the United States has access to an ECMO-capable center. This number jumps above 90 percent when effective interhospital transfer is available, under ideal circumstances. But as first responders quickly become overwhelmed during this crisis, the situation is far from ideal.

So if we do, in fact, intend to put ECMO into widespread use here in the United States as the inevitable peak of covid-19 looms, there are some important questions to ask. Can we actually offer it widely without crippling the health-care system further? Is the benefit worth the potential negative effect of limiting our ability to get other resources, like oxygen and ventilators, that may be more effective at saving a greater number of lives?

We need more high-quality data fast. We must lean on other hard-hit countries with experience using ECMO in covid-19 patients, such as China and Italy, to get this data quickly. International data sharing has never been more important. We also need to develop clear guidelines for hospitals on the judicious use of ECMO for very select patients, as the benefit needs to be carefully weighed against the risk. In these trying times, selection is critical.

We should also consider how we can provide high-quality ECMO to the greatest number of patients. Some hospital centers that can do the procedure have a low volume — meaning that these hospitals have limited experience with ECMO, and potentially worse outcomes. Bad ECMO isn’t helpful to anyone. Other countries, such as New Zealand, Singapore and Australia, have adopted a more centralized approach, with certain hospitals functioning as high-volume ECMO referral centers with dedicated inter-hospital retrieval teams. This may be something we need to consider as more and more hospitals get overwhelmed.

As we prepare for the onslaught of sick covid-19 patients, we should have these difficult discussions now. It’s not easy. Trust me: If it were myself or my family member, I would want to consider all options, even ECMO, no matter how futile it might be. These are difficult decisions for any family to make — so let’s have these conversations now before it’s too late.

Read more: