Sunday, March 11, 2007

A Proposal for an Open Source Design to Assemble Ventilators to Meet Pandemic Surge Demand

Clarence Graansma

In a predicted pandemic influenza outbreak, it is expected that there will be a severe shortage of ventilators. Ventilators are expensive to buy and maintain, so government organizations are stockpiling only a minimal reserve. Manual type ventilators will not be adequate for many cases. The solution to this problem may an open source design for an automated ventilator that will be adequate for the perceived need, and can be built from parts that will be available in sufficient quantities during a pandemic. A community of developers must design, and test a ventilator, and make the design freely available for individuals and healthcare organizations to build their own units in a pandemic crisis.

It is expected that in a pandemic influenza outbreak the number of people requiring a ventilator will be much greater than the number of ventilators that are available in hospitals. There are approximately 105,000 mechanical ventilators and 60,000 intensive care unit (ICU) beds in the United States. This is only 1 ventilator for every 2600 people and 1 ICU bed for every 4500 individuals. In an avian flu pandemic, it is estimated that 30% of individuals will become symptomatic and up to 50% will require ventilatory support. Using more conservative estimates, data from the H1N1 pandemic flu of 1918 suggested that 2% of people required ventilatory support. If the same is true for H5N1 (although less than what is currently estimated), a city of 1 million people will have 300,000 affected individuals and 6000 of whom will require a ventilator .

Based on these numbers, a city of 1 million people would have 385 ventilators in its hospitals. Since 80% to 100% of the stock of existing ventilators is typically already being used in the ICU units , this leaves at most only 77 ventilators available at any given time. Triage methods will be used to remove some of the people already on ventilators in order to give to people requiring ventilatory support due to the pandemic. Now it is possible that the impact of the pandemic may be considerably less than the 1918 event due to the use of vaccines and anti-viral medications. Let us assume again a very good response to medications, and we reduce the number of people requiring ventilatory support by 50%. The expected patient load would now be 3000. Pandemics do not always strike all at once, but may come in several waves such as the 1918 pandemic . The same ventilator could be used sequentially 2 or 3 times in each wave to treat pandemic victims in perhaps 3 subsequent waves. This means that each ventilator could now be used to treat 6 to 9 people. Assume we made up to 200 ventilators available via triage by removing existing chronic and elderly patients from ventilatory support and then used each of these ventilators to save 8 people from the pandemic. This would save 1600 of our 3000 patients.

It is obvious that even strict triage and with conservative assumptions of severity, we will be short of ventilators. Even if we had an unlimited supply of ventilators, we will not save everyone. Many will die even with a ventilator and good critical care. If the availability of ventilators were not the issue, the limiting factor then would be how far we can extend our critical care support system. Physicians, intensive care nurses and respiratory therapists will also be affected by the pandemic and their ability to respond may be reduced. There are also issues of availability of other supplies. A reasonable assumption of the limits of support extension would be between a factor of 2 to maybe 3. This means our city of 1 million would need to have available between 385 to 770 additional ventilators. Now these ventilators need not have every possible alarm and treatment option, but they must have enough automation so that nurses and respiratory therapists (RTs) can run them without constant intervention.

Some hospitals and organizations are stockpiling manual ventilators for such an emergency. These are either bag type manually operated ventilators, or pressure driven transport type ventilators with no alarm systems . These have the advantage of low cost, disposability and no maintenance. These devices require extensive supervision by qualified personnel, however and will not be adequate in this situation. They may be useful if appropriate automation and alarm systems could be fitted. To buy enough full function ventilators to fulfill the need is too expensive for hospitals to consider. Even if the government were to pay for enough ventilators to supply the entire country, there would not be enough centralized manufacturing capability to supply the product when it is needed. Neyman and Irvin have published an innovative method to put up to 4 patients on a single ventilator. This system requires further testing and would not have very good monitoring ability.

Proposed solution:
From the previous sections it is obvious that what is required is a reference design for a low cost, relatively reliable ventilator that can be produced in a large quantity in a relatively short time from commonly available materials that are not in short supply. The device will not require every feature and ability of existing full function ventilators, but must have the features required to properly care for Acute Respiratory Distress Syndrome (ARDS) in a pandemic situation. The device should be automated as much as possible as to enable the existing RTs to care for a large number of patients. Also the design of controls and alarms should be intuitive so that other persons can be trained to help support the devices in use. The components used to build these devices must be components that will be available during a pandemic.

The best way to engineer and distribute such a reference design would probably be based on an open source model. Existing projects to emulate and gain organizational insight could be the "One Laptop per Child" project or the various open source software projects such as the Mozilla Foundation or various Linux branches. The non-profit Architecture for Humanity ( is doing a similar thing for designs for housing to rebuild communities in the wake of natural disasters. We need to start a Pandemic Ventilator Project now.

The ventilator must be able to be built from commonly available components sourced from the industrial and instrumentation marketplace. The component specification should be standardized as much as possible. For example a good specification would be "12V solenoid actuated air valve with a minimum flow rate of 3 liters per minute" rather than "ACME solenoid valve AS3506T." This will allow substitution if required.

A centralized listing would be have to be established and maintained of possible components that will satisfy the requirements and known supply sources. The design should incorporate "fail-safe" design techniques as much as possible. In order to use "off the shelf" components, the design will have to rely on an electronic control system to enhance safety and usability instead of using innovative pneumatic component designs. It will probably be either PLC based or some type of dedicated PC control.

Testing criteria and minimum performance specifications will have to be developed.
It is expected that alpha, beta and release candidate versions will be released and then tested. There may be version upgrades based on testing results. It may be beneficial to fork the project at some point in order utilize differing design philosophies or to produce devices tailored to certain requirements, such as simplicity of operation, desired features or ease of assembly.

A community of developers will need to be established. It is doubtful that existing ventilator manufacturers will participate on a formal level due to competitive and legal obstacles, however it is expected that they may allow some of their engineering staff to participate on their own as a humanitarian gesture. It would be expected that professional groups may encourage their members to support the project. It would be very helpful to obtain the support of university engineering labs. It is expected that the bulk of support would be individuals from the medical, instrumentation and information technology communities.

A legal framework will have to be established to protect contributors to the project from legal liability of any misuse of the reference design or any lawsuits from failure of a device. Something like the GPL will have to be used to control derivative use of the reference design. As it is unlikely that the design will be submitted for FDA approval, there would have to be legislation enacted by governments in a crises to permit use of any devices produced. Perhaps some draft documentation to guide the government agencies at the time of a crisis could be produced ahead of time. Humanitarian groups may wish to use the designs for third world relief projects.

A foundation may have to be established to support the project. A core group will have to be established to control and maintain the direction of the project. Training, servicing and operation guidelines and materials must also be produced and maintained. A website for feedback, communication and software distribution will be required. Perhaps Sourceforge could be used.

Mar 11 2007


1 Disaster Medicine: Understanding the Threat and Minimizing the Effects
Christopher J. Lettieri, MD
Medscape Emergency Medicine. 2006;1(1) ©2006 Medscape
Posted 05/31/2006

2 CMAJ • November 21, 2006 • 175(11) | 1377
Christian, Michael D. et al

3 Mass Critical Care with Scarce Resources: A Community Planning Guide AHRQ Publication No. 07-0001 Marc Roberts PHD et al

4 Taubenberger JK, Morens DM. 1918 influenza: the mother of all pandemics. Emerg Infect Dis [serial on the Internet]. 2006 Jan [date cited]. Available from

Volume 4, Number 4, 2006
The Prospect of Using Alternative Medical Care
Facilities in an Influenza Pandemic

6 A Single Ventilator for Multiple Simulated Patients to Meet Disaster Surge
Greg Neyman, MD, Charlene Babcock Irvin, MD

No comments: