Put the Cheese in the Refrigerator

Or, reducing patient harm by eliminating the sources of error.

By: Mitch Valdmanis, Senior System Designer

I just read an article in the latest issue Healthcare Executive magazine by Dr. Erin S. DuPree from the Joint Commission Center for Transforming Healthcare. In the article, titled “High Reliability: The Path to Zero Harm“, Dr. DuPree discusses achieving The Joint Commission’s goal of zero patient harm by transforming healthcare organizations into high-reliability organizations. She writes, “High-reliability organizations prize identification of errors and close calls for the lessons that can be extracted from a careful analysis of what occurred before these events. These lessons often point to specific weaknesses in safety protocols or procedures that can be remedied to reduce the risk of future failures.” The article then goes on to describe how high-reliability organizations work to eliminate patient harm though a culture of safety and blameless error reporting and investigation so that systematic changes can be made to prevent future patient harm.

This article, and others like it, make the important case that errors are unavoidable in healthcare and that we must recognize this and work to eliminate harm caused by errors. However, the fact that we cannot eliminate all errors does not mean that no errors can be eliminated at the source. This critical distinction is not made clear, to the detriment of the cause.

In a patient safety handout titled “Learning from error“, which largely echoes the guidance provided by Dr. DuPree, the World Health Organization has broken down errors into two broad types, each with two subtypes:

1. Errors of execution, where the intended action is correct but it does not go as intended. This is broken down into:
   1. Observable errors, “slips”, such as performing the wrong action.
   2. Unobservable errors, “lapses”, such as failing to perform an action.
2. Failures of planning, where the intended action is itself incorrect. This is broken down into:
   1. Rule-based errors, where an incorrect rule or guideline is applied to the situation.
   2. Knowledge-based errors, where the actor chooses the incorrect action.

I would like to posit a similar breakdown for error mitigations:

1. Error prevention; actions that prevent the error condition from arising. These can be broken down into:
   1. Elimination; changes to the system or process that remove the possibility of the error condition from occurring
   2. Probability reduction; changes to the system or process that reduce the likelihood of the error condition arising
2. Harm reduction; actions that prevent or reduce harm caused by errors that do arise. These can be broken down into:
   1. Catchment; changes to the system or process that identify and correct error conditions before they are able to cause harm
   2. Quarantine; changes to the system or process that respond to harm in an effort to minimize the spread of harm.

Considering the classic “Swiss-cheese model” of errors, the latter three harm reduction methods add additional slices of cheese (arguably, quarantine isn’t even that much), but eliminating the error risk entirely is like putting the cheese inside of the refrigerator – even if all of the holes of the cheese are aligned, the refrigerator door blocks the hazard before it even gets close to the cheese. One example of this type of error elimination in the healthcare space is the change to non-interchangeable gas connectors on anesthesia machines. Once the change was made, it became impossible to connect the gas lines to the incorrect sources. Today, anesthesia machines that lack this critical safety feature are considered obsolete.

The sources cited above focus on the second category of actions – harm reduction – and indicate that a culture of safety should further work to reduce the probability of errors, however they give no consideration to the fact that some categories of errors can be completely eliminated, despite the great success of this approach when it is available. Dr. DuPree adds, “High-reliability organizations focus on prevention, not reaction; the further upstream the problem is, the more important it is to fix.” The corollary to this is that the further upstream the fix is applied, the greater its impact, and by definition you cannot go further up a stream than to its source. The charitable explanation for the omission is that it is assumed that if such measures are available, of course they would be taken.

There is at least one still-common practice in IV compounding that is ripe for this kind of error elimination: the syringe pull-back method of checking the dose put into infusion containers. This method of dose verification has been shown to be unreliable, yet it is still practiced today. The practice, and all of its attendant error risks, can be eliminated through the use of gravimetric dose verification. Gravimetric verification ensures that the correct amount of additive is actually put into the container, without relying on any proxies.

In closing, it is heartening to see that the medical harm reduction movement is gaining momentum, particularly in the field of medication safety, but let’s also remember to keep our eyes open for opportunities to completely eliminate sources of error and put these cheese in the refrigerator as soon as possible.