This entry is part 1 of 43 in the series Words

The Whorf-Sapir hypothesis says that our language shapes our thoughts. (Recently, there was a segment on NPR news about how important dinnertime was for kids, mostly because the discussion is a great place to learn new words and concepts.)

So here’s a vocabulary word to use and contemplate as you think about ED systems: “brittle.”

When I hear the word, I think of peanut brittle; of shiny cast metal faucets that break, and the dull, greyish metal exposed in the break (“pot metal,” the plumbers call it). What do you think of?

My dictionary says:

brit·tle, adj., -tler, -tlest, n., v., -tled, -tling.


1.having hardness and rigidity but little tensile strength; breaking readily with a comparatively smooth fracture, as glass.

2.easily damaged or destroyed; fragile; frail: a brittle marriage.

3.lacking warmth, sensitivity, or compassion; aloof; self-centered: a self-possessed, cool, and rather brittle person.

4.having a sharp, tense quality: a brittle tone of voice.

5.unstable or impermanent; evanescent.


6.a confection of melted sugar, usually with nuts, brittle when cooled: peanut brittle.

–v.i. be or become brittle; crumble.

So, let’s apply this to ED systems. A brittle ED system is one that accepts pressure up to a point, and then suddenly and catastrophically fails.

What kind of pressure?

First, let’s consider the pressure of unscheduled patient arrivals. Consider a multi-casualty incident. Or a true catastrophic disaster. Or, given the way things are going (see the IOM report on the state of emergency care) – even just a usual day in a busy, overcrowded ED. If you’ve worked in an ED in the past decade, I’m sure you know what I mean. If you haven’t, sign up for some volunteer/observer time at a big hospital ED on a Friday or Saturday overnight, or on Monday evening. Or go to a military field hospital in a war zone, it’s pretty much the same.

The patients are arriving faster than the system can accept them. (Note that I haven’t said anything about computers yet, we’re just talking about ED systems, which can include paper-based systems or greaseboard systems.)

Part of the problem in such situations is that there aren’t enough doctors and nurses working in the ED. There are chronic shortages of nurses in EDs around the country, and emergency medicine residency programs don’t turn out enough graduates to even come close fill all the emergency physician positions.

Part of the problem is that there aren’t enough treatment spaces in which to see the patients.

Part of the problem is that hospital administrations like to keep admitted patients in the ED when inpatient beds are tight, to concentrate the badness in one area to protect the rest of the hospital.

But part of the problem is that the ED systems – tracking and charting systems, for example – have difficulty keeping up with the load.

Let’s take a look at ED nurse charting first. I don’t know if you’ve had this experience, but in some EDs, it takes 10-15 minutes for a nurse to “triage” a patient. Even if the ED is empty and there is a doctor standing at the bedside, ready to examine and treat the patient. Why? Well, the hospital, based on pressures from various sources, has added to the nurse “triage” assessment such things as:

  • collecting and verifying a list of all medications (“medication reconciliation”)
  • screening the patient for potential domestic violence

Now, if the nurses are using a paper-based system, and they get overwhelmed with patient arrivals, the system can bend without breaking. What do the nurses do? They simply quit getting a complete med list, and they simply quit asking about domestic violence. It’s against the rules, and they might get yelled at later on, but they can keep the patients moving. The system bends, but it doesn’t break.

As a side comment, do you know how teamsters or air-traffic controllers do a work slowdown? They follow all the rules. Truckers come to a complete stop at all stop signs. They obey all speed limits. (This is discussed in Donald Norman‘s 1998 book The Invisible Computer: Why Good Products Can Fail, the Personal Computer Is So Complex, and Information Appliances Are the Solution. Highly recommended.

If you find this intriguing, you might branch off to the study of human error analysis, and in particular, look at the work of James Reason, especially his book Human Error, where you will learn how most “human error” is actually when a human finally, after thousands of instances of coping with a broken system by bending some of the rules, fails to bend the system enough to prevent an error.

But let’s say the ED has transitioned to an electronic nurse-charting system. Let’s assume further – and unlike the physicist’s recipe for fried chicken that starts “first, assume a spherical chicken” this is reasonable assumption – that the computer-based nurse charting system absolutely demands certain things. It demands that nurses verify all medications and answer questions about domestic violence before the nurse can chart, or even put the patient in the tracking display.

Is this a “brittle” system? You bet it is. When overwhelmed, what can the nurses do?

  1. Do everything they way they were instructed to do and let the waiting room fill with hundreds of patients.
  2. Use the computer system to lie, and to simply enter “no meds” and “no DV” for all patients.
  3. Abandon the computer system and go back to the “downtime” paper-based triage charting system – assuming that they can find the paper forms somewhere.

gedankenexperiment (“thought experiment,” a term popularized by Albert Einstein), ask any of us who worked at hospitals along the Gulfcoast or in New Orleans during and after Hurricane Katrina in 2005, or during the four-hurricane season in Florida in 2004.

Another example would be point-and-click ED physician charting systems that absolutely demand that you answer a certain number of questions before you finish a chart.

So, do we want ED systems that are brittle? Or do we want systems that can fail gracefully and gradually, and then gradually come back up to operating full speed?

Proper disaster planning includes prioritizing ED system functions, making systems that gracefully move between normal mode, too-many-patients mode, and full disaster mode. One can imagine a slider bar on the computer system that can be moved between these three modes (only by those responsible for patient flow, one would hope) and that would flag all charts as somewhat incomplete due to a disaster activation.

Disclosure: this idea emerged out of a presentation on IT in disasters, given jointly by myself and Dr. Jim Aiken of Charity Hospital, New Orleans, at the annual International ED Information Systems Symposium, held December 2007 in New Orleans.

Series NavigationRobustness

Tags: , , , , , , ,

This entry was posted by kconover on Tuesday, March 18th, 2008 at 9:50 am and is filed under Disaster, Tutorials . You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

Comments are closed.