Wednesday, March 25, 2020

The SANER Project

I don't usually talk about "work" that isn't part of my volunteer standards activities here.  What I do in standards is public information, and I can have my own opinions about it, so that's safe. The rest of my work is really between me and my employer, and when I blog for them, I do it on their site as I've done for the past decade and a half.  But there are times to break the rules, even my own.

The SANER Project is called that because if I couldn't do something about COVID-19, I would go crazy.  At lot of my colleagues at Audacious Inquiry feel that same way.  That's why we've started a project to work on this.  There's an implementation guide, and people, and meetings.  If you are in the same boat, become one of the people, join the meetings, and help us develop the implementation guide, test it, and even pilot it.  So, how am I breaking my rules?  This isn't a Standards project.  It WILL become one, it is using standards, but first and foremost, it's about how to rapidly deploy a solution that can become the way to address our needs in the US.

I grew up outside Philadelphia during the gratefully brief Legionnaire's disease crisis, and I remember the trauma, and the daily news cycles as a child, and my a) inability to escape the constant barrage of information, and b) complete inability to do anything about it.  It must have had an impact on me, I can still recall it vividly today as we enter the same sort of barrage around COVID.

I've been thinking about what the Health IT Community could do about COVID-19 for weeks, I listened to what people were saying, and where the attention was (or wasn't). Discussions around situational awareness have been getting louder again as this crisis began (as is usual for Public Health in the ongoing approximately triennial cycle of novel fill-in-the-blank).  Six new diseases SARS, Bird Flu (H5N1), Swine Flu, MERS, Bird Flu (H7N9), and now COVID-19 (SAR-Cov-2) in the last 17 years.

Situation awareness is something I first heard about in standards circles with the introduction of what many of us called the "Bird Flu" Use Case in ANSI/HITSP.  HITSP adopted OASIS HAVE and some HL7 V2 messages in C47 Resource Utilization Component.  HL7 and OASIS worked on a Cross-Paradigm specification.  And AHRQ took up the ball with HAvBED2.  It didn't work because it involved manual entry of data for the most part. 

The big ask from Public health are very much the same now as they have been over the past decade an a half: How are hospitals doing?  Break that down, and it's Beds (by type), Equipment and Supplies, and Staff.

The challenges our healthcare providers are having now is dealing with a crisis, and not having resources to get public health the information they need because we failed to automate it when we had the time.  And if the situation keeps repeating itself the way it has over the last decade an a half, they won't be able to address it the next time either, because nothing can be done in the middle of a crisis, and real efforts to resolve it after haven't considered the real challenges of automating the process, because once the crisis is over, public health doesn't get to take center stage (if there's ever a group of under-recognized people, it's those who work in public health... it's somewhat like what I've heard others say about the military 98% waiting, 2% sheer panic).

Bed Availability is a math problem.  Most think about it as:

Bed Availability = Total Beds - Beds In Use

But that's half the story because a bed cannot be used if the people needed aren't available to staff it.  A bed is a location of care... if the staff cannot provide the care, an empty bed is not going to help.  The real math is closer to:

Available Capacity = Total Beds - Beds in Use
Usable Capacity = f(Available Capacity, Available Staff)

And f() varies by type of bed and so many other things.

And there are other challenges.  Total Beds is not a constant for a facility, it's a fungible figure, because someone can roll out overflow beds into the ER in minutes, reconfigure an OR as a temporary ICU in hours, crank up a temporary treatment center in hours, build hospitals in days, or field treatment centers, activate resources, and people and even Health IT systems (my employer developed PULSE for this -- That came out of the "Katrina" use case).

Equipment is fungible too, ventilators have been modified in real treatment (NOTE: This is off-label use, and not a recommendation from me, just a note that it's been done by experts), and there are makers building them.

But situation awareness isn't about exact numbers, it's about close enough to make plans (kind of like software project management).  And there's an incremental path to getting there, because we can track the outputs needed (Bed Availability) back through it's dependent inputs Available Capacity, Total Beds and Beds in Use, and start generating those numbers, and eventually, as we track it, we start to derive the f() based on what we learn.  But before we get to point C, we have to go through A and B first.

There's value in reporting available capacity and beds in use, even if that doesn't get to usable capacity immediately, because someone who works in this space can take what else they know about staffing shortages and compute an approximation of f() in their head.  And Beds in Use is an important signal alone. It can be used to alert public health to changing trends.

And tracking those two and applying analytics can eventually get us to a data driven value for f(), and other analytics can improve the performance of f() even further.

And yes, I know, there's International value in what we are doing, and I certainly welcome International contributions, but for now, this project is going to focus on the US; others are welcome to contribute and use the materials it produces, but for now, it's going to be hard enough to manage 56 or so jurisdictions and public health agencies in my own country, so it's going to be focused at least for now, on US efforts.  After all, there's only so many cats one can possibly herd.  Thankfully, I'm not trying to do this alone.



Post a Comment