How to Make a Ventilator

No single device has gotten more attention during the coronavirus pandemic than the ventilator, which gives some patients with severe COVID-19 a fighting chance to stay alive. Access to a ventilator can mean the difference between life and death, and ventilator shortages in Italy and potential shortages in the U.S. and elsewhere have become one of the most critical logistical problems to solve during the pandemic.

As every level of government wrangles with procurement and distribution of the breathing support medical devices, massive corporations like General Motors and General Electric and tech industry figures like Elon Musk are jumping into the fray to try and rapidly ramp up production or design new, easy-to-produce models. Simultaneously, efforts to open-source and build DIY ventilators have flourished. The complexity of these projects ranges from essentially automating squeezing a manual resuscitation bag, like the Israeli Air Force’s AmboVent, to adaptations of existing medical devices undertaken by hospital workers. Gaming computer manufacturers have created prototype ventilators, university professors say they have retrofitted KitchenAid mixers to make ventilators, and hackers have jailbroken CPAP machines to turn them into makeshift ventilators. Enterprising doctors and nurses have found ways to use ventilators on more than one patient. Medical device manufacturer Medtronic even made a contribution to open source efforts, releasing specifications and documents for its PB 560 ventilator, a model originally manufactured in 2010.

These efforts are laudable, urgent, and unfortunately necessary. But turning a car parts factory into a medical device factory doesn’t happen overnight, and neither does scaling up from an open-source design prototyped in an R&D lab to medical-grade equipment safe to use on actual patients. Even General Electric, whose health division makes ventilators, told Motherboard that it could not feasibly alter factories in its other divisions to create more ventilators.

The many, many steps that go into making ventilators in non-pandemic times—the design, the supply chains, the bureaucracy, the maintenance—in some respects aren’t all that different from the processes underlying any industrial product made in a global economy. Where that traditional process is currently breaking down, and what open-source projects are trying as workarounds, poses some extremely difficult medical ethics questions, including whether or not to treat medical devices like any other industrial product. While there are dozens of projects trying to make ventilators, they all have major hurdles to overcome before they would actually be used for treating COVID-19 patients.

There’s a few ways to get started in the world of ventilator design and manufacture. One option before COVID-19 looked a little like this:

• Go to medical school
• Become a respiratory therapist
• After some years in the field, transition to industry (maybe even start a company)

Given current circumstances, this path may be difficult to replicate. The good news is that lots of people who took that path are still working and very much want to help. A lot of people want to help solve the ventilator crisis right now—which is great, because it’s going to take a lot of people to do it, and no one person can single-handedly take it on. In addition to the necessary collaborative approach required to weave together complex hardware and software engineering, in a hospital setting these devices also require respiratory therapists and technicians who undergo lots of training to maintain and operate them.

Most people within the field already know and understand this, and a number of the DIY approaches being tried right now recognize the necessity of collaboration with experts. Jeffrey Yoo Warren, a founder of the community science nonprofit Public Lab and longtime participant in open-source hardware projects, has been tracking some of the emerging projects and observed, “The people who are really under-recognized as innovators here are health workers themselves,” citing the work of Mt. Sinai doctors in New York City who developed a protocol for repurposing BiPAP machines as an example.

“Overall I think there's a disconnect between the ‘maker’ ideal of innovation and what makes up part of patient care on a day to day basis,” Warren continued. “A ventilator is not a solution. It's a tool that is combined with other tools or therapies in a creative and responsive way by a medical team.”

Given the intensity of demand and the flood of media coverage, it's understandable how "ventilator" as a term has been collapsed into describing a single item, rather than a class of medical equipment that comes in a variety of forms. Let’s clarify a little.

The family of medical devices that assist people who can’t fully breathe on their own includes ventilators, BiPAP and CPAP devices, and manual resuscitation bags. The differences among these lie in how many variables they can control and how precisely they can measure those variables.

When using a manual resuscitation bag, the main variable is how hard a person squeezes the bag. BiPAPs and CPAPs both use a method called positive airway pressure (that’s the PAP part). It takes normal air, compresses it, and then sends that air through a hose to a patient wearing a mask. BiPAPs can get a little more fancy. Where CPAPs just continuously move pressured air through a mask, BiPAPs have two pressure settings—one for inspiration (respiratory-speak for “inhalation”) and expiration (exhalation) . All three of these are considered “non-invasive” ventilators, meaning the patient has a mask put on their face and doesn’t have a tube inserted into their trachea for breathing assistance, which is how most of the class of ventilators currently sought by public health agencies and hospitals around the world work.

The world of invasive ventilators covers a wide spectrum of uses. There are anesthesia ventilators for people who need to go under for surgeries, portable ventilators for ambulance care, neonatal ventilators for prematurely born babies with delicate lungs, home healthcare ventilators for people who are healthy enough to be discharged from a hospital but still need breathing support. The ventilators most acutely needed for COVID-19 patients—the kind that state governments are having to enter bidding wars to get, and General Electric and Ford are scrambling to retool their assembly lines to make—are called "critical care" ventilators. These machines, which can at their lowest-grade cost about $10,000 and at their most elaborate $50,000, have to precisely deliver and monitor far more variables than CPAPs and BiPAPs. They need to be able to change air pressure, volume, flow, and rate depending on the patient’s changing condition, which means they’re heavily reliant on sensors and a microprocessor to make those calculations.

For the most part, the various open-source initiatives underway make no claims that they’re building critical-care ventilators. Govind Rajan, an anesthesiologist at UC Irvine’s medical school and a contributor to the Bridge Ventilator Consortium ventilator project, described the use-case for that project as “only in situations where you don't have any ventilators available and the patient needs a ventilator.” In collaboration with the consortium, Virgin Orbit has designed a ventilator of the “automating-a-manual-resuscitator” variety. It’s nowhere near as complex as a critical care ventilator.

However, Rajan also described scenarios where “there comes a time when you have to be weaned off a ventilator,” and said his team’s design could serve the needs of patients who need to be weaned off and don’t need a critical-care device (i.e., acting as a “bridge” between critical care needs and being off of the ventilator). On its website, Virgin Orbit also describes the ventilator (which has still not been approved by the FDA) as potentially serving “the huge volume of patients with moderate COVID-19 symptoms.”

This seemingly contradictory description—a ventilator that’s somehow both a worst-case-scenario only option and serving an intermediate stage of COVID-19 treatment—introduces a serious medical ethics question in the drive for more ventilators. For doctors trying to save patients by any means necessary, a minimum viable ventilator is better than having no ventilator at all. Rajan recalled his own experiences when he began his career working in India 35 years ago, where ventilators were often in short supply and manual resuscitation was sometimes the only option for keeping a patient breathing. Getting to choose between the last-resort tool and a critical care device is a privilege that some doctors just don’t have right now.

“Lots of things are better than nothing. The question is, is it capable of meeting the demands of the patient?” asked Rich Branson, a professor in the department of surgery at the University of Cincinnati and the editor-in-chief of the scientific journal Respiratory Care, when asked about this rationale. As doctors learn more about COVID-19 and treating it, there’s been some reassessment of racing to produce “last resort” tools over simply investing in critical care ventilators—for example, the UK’s National Health Service recently cancelled a contract for thousands of ventilators built by teams at the auto racing industry’s Formula One Group because it was deemed insufficient for meeting COVID-19 patient’s needs. Meanwhile, projects that describe themselves as suitable for patients who are infected with COVID-19 but don’t (or no longer) need a critical-care ventilator run the risk of “inventing a patient that doesn’t exist,” Branson cautions.

Branson appreciates the good intentions of these projects, but worries the public—and some of the organizations and federal agencies funding them—don’t fully understand that they are stopgaps, not a solution for not having enough critical care ventilators. “The worst solution is the solution that makes people think they're receiving effective treatment when they're not,” he said. Understanding what kind of ventilator a patient is receiving, and what it can and can’t actually do, is crucial to avoiding any misalignment of expectations.

While we’ve already established that the complexity of ventilators varies, in the context of microprocessor-equipped critical care ventilators there are a few key components to consider: pneumatic architecture, electrical and software architecture, and power supply.

The pneumatic architecture of a ventilator does the work of taking in pressurized gas (air and oxygen), raising or lowering that gas’ pressure depending on what the patient needs, moving air around, and controlling the oxygen level and flow of air to the patient. Systems for pressurizing gas vary but often, they’re powered by turbines or a built-in compressor, or from a hospital’s medical gas supply piped into the wall. The air moves through a series of valves and tubes, and controlling that flow usually happens with a solenoid valve (an electromechanically operated device that expands or contracts depending on the flow required).

The electrical side of the ventilator is perhaps a more vital piece of proprietary secret sauce—it's the system of sensors that monitors both machine and patient inspiration and expiration, and the software that uses information from those sensors to determine the pressure, volume, and flow of air a patient receives. Calculating these variables (which, depending on how fancy the ventilator is, will be recalibrated by a technician or automatically updated by the machine) is critical for a patient's safety: if the pressure or the volume of air flowing in is too high, it can injure the lungs.

While having a power supply for an electrical device might seem obvious, ventilators also need to be equipped with a robust backup power source in the event of an outage or if a patient needs to be moved. In the UK government's specification document for a Rapidly Manufactured Ventilator System issued to manufacturers, they request that ventilators have a backup battery supply of at least 20 minutes but would welcome the option of utilizing hot swappable batteries to allow for up to 2 hours of continuous use.

Ventilators, like all other medical devices, have no shortage of bureaucratic paperwork and processes they have to go through for FDA approval. Rogers and Branson each separately recounted times in their careers when FDA paperwork for ventilators had to be printed out and submitted on a pallet. Right now, that process has been expedited by the FDA to allow for manufacturers to modify previously approved ventilators and to introduce new ventilators for approval.

That being said, “If you're going to start from scratch it's going to be hard to get it on the market even with the FDA [emergency use authorization]” according to Deborah Jennings-Conner, director of Global Regulatory and Testing Assurance for the Life and Health Sciences division of UL, a global safety science company.

At UL, Jennings-Conner has worked on the development of industry standards for medical device manufacture, and she’s been working with manufacturers on implementing new protocols in their factories so they can safely transition to making medical equipment. She’s encouraging non-medical companies who want to help to adapt or modify existing, already-FDA-approved ventilators. When bringing those existing designs to a non-medical manufacturer “it's kind of like just having a new manufacturing location,” rather than starting entirely from scratch. If manufacturers can focus on “just getting staff trained and [introducing] them to the process” of medical-device manufacture, they’ll be able to move much more quickly than brand-new designs.

But to get to that production stage, first we need to have a supply chain for all of your device’s components. Ventilators have a lot of parts, and ventilator manufacturers don’t actually make most of those parts. While the schematics and components of ventilators are generally proprietary information, we can glean some insight from Medtronic’s “open sourcing” of its PB 560 ventilator—with some caveats.

While the company went to great lengths to share this material while simultaneously ramping up production, it paints a slightly incomplete supply chain picture. Medtronic released the PB 560’s documentation in a series of staggered releases, which means that there’s some variation in the level of detail across the documents. In one release, Medtronic published individualized bills of materials for six components including the names of vendors for those components; in the final release, it published a single bill of materials for the entire device without vendors. But even with those gaps, the PB 560 documents provide a greater view into the sheer complexity of ventilator supply chains.

The assembly documents, for instance, list a total of 65 different parts. The total bill of materials contains 613 different items. Among the six bills of materials for specific components published (mostly for printed circuit board components), there’s a total of 78 different suppliers providing things like capacitors, diodes, and resistors, each of whom likely have their own raw materials suppliers. Coronavirus is affecting the global supply chains for everything, so it's possible those raw components aren’t readily available. To reiterate, that’s suppliers for just a fraction of the device’s 613 items, and the PB 560 is one of the relatively simpler models of ventilator available.

“You'll find a lot of ventilator manufacturers sourcing components from the same places,” said Mark Rogers, a project manager at ventilator manufacturer Nihon Kohden OrangeMed. Rogers has been in respiratory care for over 35 years, and worked on the development and rollout of several critical-care and neonatal ventilators. Companies looking to ramp up ventilator production right now are competing over the same supply of parts. “It's a very incestuous business. People move between companies and you start seeing them using the same concepts or suppliers,” he said.

This contributes to the challenge of having enough ventilators at the ready, though different components face different challenges. Some of the more conceptually central components like a turbine used to pressurize gas will typically be a bit more bespoke, so there’s less competition for those parts—but a lot of the constituent parts of the ventilator like gas fittings or hoses or solenoid valves are currently in short supply.

Depending on the approach being used, DIY and open-source projects will face more or less of the same challenges. The Bridge Ventilator Consortium, for their part, deliberately sought out components outside of the traditional medical supply chain and thus in lower demand, such as windshield wiper motors.

Rogers expressed some frustration at the way supply chain logic works against emergency medical preparedness. “You can't really do just-in-time supply chains for equipment like this,” he said. “We're not making paper clips. We're making life support equipment that will sometimes have unexpected surge requirements.”

Part of the challenge of ventilator supply right now is due to the fact they have been beholden to just-in-time supply chains for so long.

“This exact same problem has been described before—it's not like we've never considered this,” Branson said when I asked him about his past research on respiratory care and disaster operations.

In 2014, he co-authored a paper in the scientific journal CHEST in which a series of key recommendations were issued for addressing critical care needs in the event of a capacity surge, including the need for coordination of mechanical ventilator access across state and city hospitals, and a 3-page table outlining baseline operating, performance, safety, and maintenance features for selecting stockpiled mechanical ventilators. According to Branson, the current stopgap and DIY projects being proposed can’t meet those requirements.

As far back as 2003 and as recently as 2017, the Government Accountability Office and the Pentagon issued reports on the importance of having an adequate supply of well-maintained ventilators and personal protective equipment for healthcare workers in the event of a pandemic. What efforts have been made to fill that gap, before and after COVID-19, have been stymied in part by incompetence and in part by capitalism. As the New York Times reported last week, efforts to build low-cost ventilators explicitly for the National Stockpile starting in 2007 collapsed amid the early aughts period of mass healthcare industry acquisitions and consolidations, where a government contract to build low-cost ventilators didn’t really look like a big revenue generator. Thousands of ventilators in the National Stockpile don’t actually work because a contract lapse meant they weren’t receiving regular maintenance. It took weeks and 5,000 deaths across the country for Donald Trump to actually utilize the Defense Production Act to support increased manufacturing.

Worldwide, there’s also the problem of critical care ventilators simply being expensive and hard to make. Here, there’s fraught potential for DIY projects. Rajan from the Bridge Ventilator Consortium mentioned that his team’s been contacted by doctors in India and South Africa–places where, as Rajan put it, "they don't have a GM" to enter into large-scale production of critical care ventilators. Rajan hopes that his work can help in the short-term, but acknowledged that ultimately more systemic change is needed. In a crisis, it’s great that these better-than-nothing devices made with off-the-shelf and more easily available parts might be able to make a difference in poorer countries with fewer critical care ventilators. But giving poorer countries the tools of last resort–rather than making it easier for them to get access to critical care ventilators, whether through funding or manufacturing support–runs the risk of making what should be a stopgap into yet another unequal status quo.

Which is a shame, because respiratory care isn’t really an industry people get into because they crave celebrity or huge paychecks. They do it because it matters, and they do it because they want to bring people the best care possible.

“These are not commodities," Mark Rogers told me. "Ventilators that do special things are what people such as myself live for.” For him, the devices he’s worked on are his life’s legacy. Imagine the kind of legacy he could leave, that all of the people who’ve dedicated their lives to ventilator design could leave if access to their designs wasn’t limited by supply chains, by income, or by the treatment of healthcare as yet another commodity among many.