The treatment of last resort for Covid-19 victims with breathing problems is an invasive ventilator. It’s a horrible treatment, and only a fraction of those placed on it survive, but usually people only go on it if they would suffocate without it. In places that are having the worst outbreaks, there is a shortage of ventilators. In Italy, there are reports of people being sent home — to probably die — because there was not a ventilator and other needed care for them. When times were very uncertain, some forecasts suggested that as many as 900,000 ventilators would be needed — with only around 200,000 available in current supply. Globally, that would predict a need for millions of emergency ventilators.
To meet that shortfall, most ventilator companies started ramping up production as high as they could. Unfortunately, these devices are expensive and specialized and usually sold in fairly small volumes. As such, it was a big stretch for many companies to tool up to produce as many as 10,000 per month. That’s a lot for ventilators, but a drop in the bucket for what was predicted.
Surprisingly, it was automotive companies that stepped up, or were called upon, to try to solve this. They know a lot about high volume manufacturing of complex devices, though not hospital grade devices. Companies like Mercedes AMG (a racing division) started producing a simpler CPAP style device. General Motors GM made plans to use one of its mothballed production lines, then were “ordered” by President Trump to produce ventilators under the Defence Production Act, in concert with a ventilator company. (It is unclear if they actually had to be ordered, or if this was just a publicity stunt for the DPA.)
TSLA then got into the game coming up with their own new ventilator design which, rather than being an existing design, made use of as many parts from Tesla cars as possible, because the engineers there are familiar with them and know they can get them in large quantities. Tesla’s approach has been praised and criticized, but it opens up some other questions.
All these efforts have talked about making ventilators in the tens of thousands. Each one is a potential saved life or two, but a drop in the predicted bucket.
The real challenge would be to find a way to, in just a month, produce several hundred thousand life-critical devices. This means a design that is well tested, and a manufacturing process that’s also mature, fast and reliable. That’s a tall order.
The hacking community also heard the call, and dozens of projects appeared to find ways to produce ventilators easily, at low cost, and with commonly available parts. Many of these projects worked by taking well known manual ventilators, sometimes known as bag valve masks or “ambu bags.” These simple ventilators just use an air bladder that an emergency first responder can squeeze with their hands to force air into the lungs of the patient. You can do basic ventilation of a patient this way, but a human has to sit there squeezing the bag 24/7. Many teams built devices to have motors squeeze that bag. Since the bag is an approved ventilator, and cheap, it is hoped that is a quick path to a suitable device.
Another approach has been to use the parts from CPAP machines. CPAP machines contain many of the same elements found in ventilators, including computer controlled blowers, pressure sensors, humidification and sometimes more, such as oxygen saturation inputs. CPAP is used to treat sleep apnea, a condition which affects a very large population, where the airway closes up during sleep, temporarily interrupting breathing. By increasing the pressure in the airway and lungs, it can be kept open, giving patients a night of clear sleep. The pressure is constant, and ranges from 5cm to 20cm of H20. (This means that the pressure would keep a column of water that high above normal. 20cm, the maximum CPAP pressure, is only 1/4 psi, or about a 2% increase over ordinary atmospheric pressure, so it’s not a lot, but this is quite a lot for the lungs.) A cousin of the CPAP machine is the BiPAP, which is a little more like a ventilator in that it uses a higher pressure when the patient breathes in, and a lower one when they breathe out, but it is not a ventilator.
CPAP and BiPAP therapy are being used on Covid-19 patients, along with some additional oxygen. Fortunately not everybody needs a ventilator. The extra pressure can help open the internal lung structures that exchange gasses like oxygen with the blood. In ARDS — the lung condition caused by Covid-19 — these fill with fluid and are blocked. You suffocate.
The full ventilator
A true ventilator will breathe for the patient who can’t. While ideally their breath cycle is triggered by an attempt to breathe by the patient, the device does the rest, increasing pressure to drive air into the lungs, then reducing it once enough air has flowed in so that exhalation can take place. In many cases, if the patient is not even trying to breathe (as happens from both disease and the heavy sedation many patients are under) the machine will initiate breath for the patient, doing the whole task.
True ventilators tend to be computer controlled. They monitor both the pressure and the volume of air given to the patient. They will use pressures of more than 20cm, and sometimes go as high as 40cm, but 30cm is a more typical maximum. To make it work long term, it gets ugly — they stick a tube into your lungs and seal it there, in a process called intubation. As you might guess, having such a tube down you throat and into the lungs is immensely uncomfortable and they have to sedate you fairly heavily. It’s much better if you can have non-invasive ventilation, which can be done with a mask over the face or a helmet, but usually not for the time periods needed for a serious ARDS patient.
(There are ventilators that put your whole body, except your head in a chamber, and use negative pressure to push regular air into your lungs. The most famous were the iron lungs which saved many polio patients. These are bulky and expensive and largely gone, though they might actually be better for Covid patients. Some of their successors are in use, and some hacking teams have tried to make them.)
You don’t want to go on that full ventilator. It’s the 2nd worse thing that can happen to you, some would say, the worst thing being suffocating because you don’t go on it. Even so, the survival rate for those going on them long term is poor, and even worse for Covid-19 patients. Around the world, between 70% and 85% of patients who go on them never recover. Doctors are very eager for alternatives, but still want them for those patients who have no other choice.
From hacking to mass production
All the hacking and automotive projects suffer from that core problem — if you need hundreds of thousands of medical grade “bet the patient’s life” quality devices, it’s not practical to do that with new designs and new manufacturing lines. Research showed the blowers in the CPAP machines were able to produce 40cm or more, and so the question arose of whether these machines could be adapted with software upgrades to perform full ventilation. They don’t have a flow sensor, which is generally required for a ventilator, but in many cases flow can be calculated from pressure readings if you know the characteristics of the machine and tubing, or you can add just that extra sensor.
The reason this has been considered is that CPAP machines are already made in very large quantities. Unlike ventilators, where making 10,000/month is an emergency project, one CPAP vendor, ResMed, sells around 2.5 million CPAP units a year. Other companies sell millions more. The major CPAP vendors, such as ResMed, Philips Respironics and Medtronic Puritan Bennett, also make ventilators and are highly experienced at the requirements of these devices and have software to control the ventilation cycle already in place.
After investigating this question, I spoke to Carlos Nunez, chief medical officer for ResMed. “We get hundreds of people asking that question every day,” reported Nunez, saying that the company was definitely aware of the issues and working on it.
The good news, though, is that the predicted vast ventilator shortage has not appeared. It happened for a while in Italy, but since then, the need has fortunately been lower. “No one is asking us, hey, can we take home CPAP machines and convert them into ventilators,” said Nunez, because they are not as desperate as we feared for more ventilators. While there are not enough full ventilators to go around in New York, success was found moving patients to BiPAP therapies and simpler ventilators. The 1,000 machines Elon Musk donated were actually ResMed machines sourced in China according to Nunez.
Since then, California has found it has more than enough ventilators and is loaning some to other states. California has weathered the storm well, and doctors have been working very hard to find alternatives to ventilating patients.
My own investigations pondered whether there were millions of CPAP machines already present, mostly unused, in people’s homes, which might get converted with new software and small amounts of additional hardware into ventilators. Compared to new designs, these systems would be already manufactured and tested. While millions purchase CPAP machines, many find they don’t tolerate the therapy well and give up, leaving the machine on the shelf. Others upgrade machines and store the old one. It is suspected that a large number of people might gladly offer the unused machines, or go back to their old machine to let their new one save a life. (Many also often sleep without their machine, at a minor risk to their own health, which they would gladly take if they could save a dying patient by doing so.) According to Nunez, the number of spare machines is not nearly that high, since patients who give up on the therapy usually are made to return the machine if it was paid for by insurance.
For now, the goal of companies like ResMed is to prioritize, and make sure they can get as many of the right types of machines out to the hospitals that need them.
In spite of the reduced need, many have taken action. Doctors at Mt Sinai hospital in New York converted 500 donated BiPAP machines to basic ventilators to use on patients who did not need a full hospital grade device. Doctors and engineers at UC San Francisco and UC Berkeley solicited donations of machines and received around 1,000. We’ve seen doctors in Germany convert traditional CPAP to ventilators with the addition of a servo controlled valve to alter the pressure and flow. A team at Auburn University did their own adaptation.
Traditional BiPAP gear, including the mask that goes over the patient’s nose or face, vents the exhaled air directly out. This presents a risk of spreading active virus particles. Ventilators more commonly have a means to filter exhaled air, or even recirculate it after removing CO2.
Another problem is oxygen. ARDS patients all need additional oxygen, and the most severe cases going on ventilators often start at almost pure oxygen, though that can’t be tolerated for very long. Putting pure oxygen through hardware not designed for it carries risks — oxygen leaks into areas with hot electronics could trigger fires. ResMed’s testing has shown their equipment can handle “pretty high” concentrations of oxygen before there’s a risk.
Another interesting approach is a $100 simple ventilator made by Vortran. Their Go2Vent operates using the high pressure oxygen available in most hospital rooms and without any electricity. Like many of the simple designs, it does not do the fancy computer controlled monitoring of flow and pressure, but it’s simple and FDA approved.
The FDA has expressed some willingness to relax standards for ventilators designed for the emergency. After all, if a patient faces the choice of taking a less tested emergency ventilator or suffocating, the choice is not very hard. Fortunately not too many are being forced into that choice at present. Since the epidemic is hitting peak in different places at different times, those places that are past their peak are willing to loan equipment to those going into it. The United Kingdom published standards on what an emergency ventilator must do, and the FDA has been working with companies to develop US versions, and has some preliminary regs. (ResMed told me they are working with the FDA on a document to cover this situation.)
While CPAP companies have not felt the need to make new firmware for their machines yet, others have reverse engineered the products to produce their own “jailbreak” firmware for existing machines. The AirBreak project modified the very popular ResMed Airsense 10 allow it to be a BiPAP and even operate in a low-grade non-intubated ventilation mode they call iVAPS. This is possible because today, many manufacturers use the same hardware in all levels of a product, from the low cost to the expensive, and enable the more expensive features in software. It is cheaper to mass produce one hardware design. In addition, with computer controlled blowers, the hardware is not that different. AirBreak still thinks more testing is needed before using their hack to help patients, and ResMed declined to comment on it.
Let’s hope that we never need a massive supply of emergency ventilators. If that need comes, it seems as though the CPAP companies will be willing to release software enabling more function in their machines, so long as they can get the cooperation of the FDA and other federal agencies to make it legal and shield them from the liability that could come in using machines beyond their intended purpose. Hopefully that cooperation would not be too hard to get. Nunez says he gives “tremendous props” to the FDA for the speed at which it has been working. Not only can machines be adapted (or given simple additional hardware) but there are ways to introduce oxygen after the machine to provide up to 50% oxygen, and also ways to modify patient interfaces (masks and tubes) to filter the output and protect those around patients. If the world needs it, the supply of machines can be there. Sadly for the ventilator companies, the massive ramp-up in production will be followed by a serious sales drought due to the oversupply that will exist at hospitals.
Sadly, the demand is not yet known in Latin America and Africa. It may turn out to match some of the frightening predictions.
Those who wish to go deeper can read my earlier exploration of this topic which contains links to some other projects.