The Smart Pumps That Cried Wolf​​

January 9, 2017 | Sean O'Neill

Wolf face photo

We all know the childhood story about the boy who cried wolf. The story, derived from Aesop’s Fables, is really quite simple: A boy repeatedly enters his village screaming that wolves are attacking a flock of sheep. Of course, he’s telling a “tall tale” and—after repeatedly doing so—the villagers no longer take him seriously. When the wolves actually enter the village, the boy alerts the villagers to no avail and the sheep are eaten.

So how does this relate to healthcare and patient safety? Are US hospital becoming overrun with wolves? Not literally, of course, but this phenomenon plays itself out in our hospitals every day. Clinicians interact and utilize tens, if not hundreds, of different technologies in the acute care space. These technologies have the capabilities to prevent patient harm by alerting clinicians when they are about to effectuate an unsafe practice. However, if the alert parameters set on the technology do not match practice, the result can be an overabundance of nuisance alerts.

The Agency for Healthcare Research and Quality (AHRQ) states that alert fatigue describes the phenomenon by which busy workers (in the case of health care, clinicians) become desensitized to safety alerts, and as a result ignore or fail to respond appropriately to such warnings. In 2015, The Joint Commission made alarm fatigue a National Patient Safety goal and now requires hospitals to make alarm and alert management a hospital priority.

The use of technology with safety alerts is very evident in the ordering, preparation, and administration of medications in an acute care hospital. What most non-clinicians, or many clinicians for that matter, don’t know is that depending on the type and form of a drug, it is possible that the medication process will include up to five or more independent technologies supporting a medication along its path to being administered to a patient.

Let’s breakdown one example of the medication process for an intravenous medication. In most acute care settings, this medication will be ordered via an electronic health record or computerized order entry system (technology #1). The order is then transmitted to the pharmacy for review and preparation. Within the pharmacy multiple storage and preparation technologies may be used including medication inventory control systems (technology #2), IV robots (technology #3a) and an IV workflow manager system (technology #3b). Once the preparation is complete the medication can be delivered to an automated dispensing cabinet (technology #4) for storage in a patient care area. Prior to administration to the patient, the nurse will use bar code scanning (technology #5) to verify four of the five rights of medication administration. Finally, the medication is infused via a smart infusion pump (technology #6). As you can see, this couldn’t be kept to less than five different technologies! Keep in mind the above process happens hundreds of times per day in any given hospital.

Why make the system so complex? All these technologies have the potential to provide a more efficient and safer administration of potentially high risk medications. However, there are two major limitations in optimizing the use of these technologies. This first is that they require the development and maintenance of the clinical content that drives their value. For example, smart infusion pumps can alert a clinician if they program a 50 mg dose vs an intended 5 mg dose for a given medication. Unfortunately, there is no standard for which drugs should be included in your library or at what value your alert thresholds should be set at. Therefore, hospitals end up having to create this content from scratch and often in isolation from other healthcare organizations.

The second major challenge is that these technologies all produce transactional data that can be used to identify vulnerabilities or deficiencies in a hospital’s processes. If harnessed correctly, this information could drive improvements and ultimately make patients safer. This second challenge is where I want to focus.

You may be asking how this all relates to the boy who cried wolf? The answer is that all day, every day, the above technologies are crying wolf. In theory, all of these system have alerting technologies that can halt a clinician from making an error. In 2016, the ECRI institute identified alert or alarm fatigue as a Top 10 health technology hazard.

To further illustrate this, let’s focus on smart infusion pumps. For the clinicians reading this, I have a homework assignment. Go to one of your patient care floors or operating rooms and find a nurse or anesthesiologist. Ask them the following question, “Do our infusion pumps alert you to potential dosing errors?” I would bet a large sum of money that their answer will be “Yes, I see the alerts but they never make sense so I ignore them”.

How do we address alert fatigue in this setting? The vital steps consist of having access to the data and translating that data into insight. You can’t optimize alerts or alarms in any given system without first understanding the incidence, type and action taken with alerts. If we go back to the smart pump example, can your organization answer the following questions:

  • What is the alert rate for all the drugs in our library?

  • What are the types of alerts that are firing?

  • What values are triggering these alerts?

  • What are clinicians doing when faced with these alerts?

In talking to countless medication safety leaders across the country, you would be surprised how few organizations can answer these types of questions about even one or two of the technologies they utilize. And the ones that can are spending a significant amount of resources to do it.

This is where Bainbridge Health comes in. We provide a software data platform as well as clinical consultation to help healthcare organizations optimize the use of their data and translate it into clinically actionable information. It is our goal that technologies will no longer be crying wolf but instead will only alert clinicians when the wolves have actually entered the village.