Rethinking Disposables

One of the core motivations for our work at Mechanomy is the belief that many of today’s systems are too complex.  While complexity is inherent in every system, a significant portion is incidental to the core problem, added to the system by inefficient business, development, and production processes.
And it is this incidental complexity that the ongoing COVID2019 pandemic is particularly revealing: anecdotes abound of situations that, while defensible in normal times, appear unwise under today’s more trying circumstances.

Take, for instance, CDC’s notice that some ‘expired’ N95 masks remain usable.  While only the manufacturer and the CDC know what actually limits the usability of a mask, it is likely not the critical element of the filter, but rather that the rubber gasket becomes less soft with age and seals less effectively.  Sealing is a critical aspect of a mask, but it is one that might be remedied by tightening the face straps, etc.

When we look at the requirement to wear masks, then, we see that the requirement’s sensitivity is not expressed, modeled, or known (publicly). We don’t appear to know how the masks’ effectiveness degrades with time: I would expect that, when new, a N95 mask (95% of particles are filtered out) actually filters, say, 98.5% of particles and that the 95% is only reached at the designated expiration date (the design life of the mask).  That’s a slow degradation curve, 3.5% reduction over, say, 3 years; if the alternative is no mask at all, that expired mask is still useful.

These are my conjectures; who actually knows this?  Is this kind of information communicated on the product packaging or is it withheld so as to guarantee compliance with the manufacturer’s directions and the maintenance of a particular standard-of-care?

It is simply amazing how many single-use disposable implements are employed in modern medicine.  We have an open-loop system (meaning that it is trust-based, that there is no measurement of the result): implements are produced with a particular quality and as long as their packaging is intact, they’re stored correctly, and they are used within their stated lifetime, they will remain in the expected or usable condition.

This pandemic is showing two key weaknesses of this usage model.  First, the end-user is not able to verify the integrity of the implement; normally this is not an issue but when the supply is threatened the user cannot determine when something has expired.  In the case of masks, there are reports of practitioners wearing disposable masks for multiple days, yet certainly the mask was designed for only x hours of use, meaning approximately x*b breaths and a total flow of x*b*f through the filter.  As more flow through the filter will tend to erode the filtering medium, creating larger air channels, the mask’s effectiveness should be expected to degrade…but what is this limit?  CDC has some guidelines, but these do not appear exact.  Instead of relying on infinite resupply, it would be really cool to have a smell-based mask-checker:  the idea being to introduce an odorous substance into a controlled volume, so that if the mask wearer smelt the substance they would know that their mask is compromised.  Filtration is more complex than this, but many odors are available [PDF]. In preferring an open-loop, disposal-first usage model, we have avoided developing or fielding mask verification tools.

The second weakness of the disposal model is that it is entirely opposed to reuse. In the case of COVID, how long do the particles remain active in a mask?  Is it possible to clean the mask without damaging it or to wait, say, 7 days until the substances have died and then re-wear?  When will the non-COVID contaminants become inert?  Again, the disposal-first usage assumption resists this question and course of action.

When regulating the function of various systems, some regulations are essential to the proper use of the system — how to adjust the mask to your face — while others address externalities — a new mask shall be used for each unique patient — whose relevance can only be known by those in a particular situation. This demonstrates why it is important to model requirements, and particularly their sensitivity to various forcing functions in the course of designing a system AND to express that modeling to the end user so that they can understand when the context has changed.  Inasmuch as the manufacturer’s intended use case motivates the original requirements, truly robust designs tolerate off-nominal conditions in predictable ways.

Do hospitals and regulatory agencies model their requirements?  Do they understand how these requirements constitute their standard-of-care and how do they determine when particular conditions can be relaxed to accommodate demanding situations?  What conversations can the systems engineering community have with the healthcare industry to build more known and robust health systems?

Our mission at Mechanomy is to create known systems; while most of our time is spent in mechanical systems, our tools and approaches generalize to any complex system.  We’re happy to chat if you want to know more.  Also, I would be very interested to hear if anyone has any anecdotes, @mechanomy_ or ben @ mechanomy.com. Thanks to all on the front lines and, to everyone else, keep your distance!

Leave a Reply

Your email address will not be published. Required fields are marked *