Alert Fatigue: The Pitfall of Clinical Decision Support?
Prior to the advent of Computerized Provider Order Entry (CPOE), orders for medications and IVs were either faxed or copies were sent to the pharmacy department allowing the pharmacist to enter these orders into the pharmacy computer systems for label processing, charging and maintenance of a patient’s medication profile. The pharmacy system usually had some type of clinical decision support available to alert the pharmacist of potential problems with the medications they were entering, should any be present. These reminders or alerts would pop-up during the medication order entry process and the pharmacist would act accordingly by either making a note on the order to call the provider or overriding the alert because it was inconsequential. Essentially, the workflow in place was a screening process in which the provider would only be notified about an alert if a pharmacist, based on their discretion, felt it was absolutely warranted and necessary.
So when discussing clinical decision support, which alerts are we talking about? There are: drug-allergy, drug-drug interaction, drug disease interactions, duplicate drug warnings and drug dose range reminders, however this is not an all-inclusive list as these types of alerts are numerous.
After the introduction of CPOE, alerts found a new home in the order entry process that the providers were performing, allowing for the providers themselves to be alerted as they were entering medications and/or IV orders. With this shift in workflow, it is not surprising that in today’s day and age, the most common phrase heard when discussing clinical decision support is alert fatigue. This phrase refers to the process in which an ordering provider receives an abundance of pop-ups or alerts during the order entry process many of which are clinically inconsequential, causing the provider to be largely desensitized to these notifications. This can result in the alerts losing their significance, which can in turn result in a provider either just overriding the plethora of alerts present or choosing to just simply ignore them.
Ideally these alerts should occur when the ordering provider is placing the order for a new medication. No one would argue that reminding a provider of a patient’s allergy to a specific medication being ordered is not a good idea. The issue that can arise is when a provider is ordering multiple drugs for one patient and the prompt for that allergy warning is just one of many instances of warnings that pop-up during the process. The screening process in place performed by a pharmacist seen in the pre-CPOE workflow is no longer in place, therefore it is not out of the realm of possibility that more often than not a truly significant warning or alert could be sandwiched between a dozen or so clinically inconsequential alerts. This is seen more and more as Electronic Health Systems are set up for alerts to occur as a batch process in an attempt to streamline the ordering process for providers and minimize the number of clicks they have to process.
It is safe to say that CPOE is not going anywhere anytime soon, and because of this it is critical that alerts are applied judicially to ensure first and foremost that the provider receives pertinent information and is not overly flagged during the process. One way to accomplish this is to minimize the alerts just to those that are serious or necessary for the provider to review. This can be easier said than done and requires an informed analysis and agreed up on workflow for each type of alert; this post will only discuss a few.
Drug to drug interactions alerts can typically be set up so that an ordering provider is flagged for severe drug to drug interactions only. A similar approach of necessity can be taken for drug to disease type of alerts, in which there are many. If a patient has hypertension, it is well known that there are drugs that they should avoid taking due to the fact that they will most likely increase their blood pressure. In some cases in this scenario these alerts are an absolute and therefore warrant the provider being warned, but in others the appropriate alert is simply one that the patient should be monitored closely for this side effect. This begs the question, is it necessary that the provider be subjected to all the possible drug to disease types of alerts regardless of the significance?
Duplicate therapy reminders and drug dose range reminders are two types of alerts that are more complex to navigate through. Duplicate therapy alerts occur when two or more drugs in the same class or similar classes are ordered at the same time. For example, in the treatment of pain multiple opioid containing medications may be used simultaneously depending on the level of pain treatment. The same could also be said of other step wise treatments like hypertension. An order for one antihypertensive could exist on a patient but still their blood pressure is not controlled well. The provider could then order an additional different antihypertensive along with “stepping up” therapy until the blood pressure is controlled. If the ordering provider receives a duplicate therapy alert(s) when placing orders for any of these types of step therapies, the provider is very possibly going to ignore these warnings due to their irrelevance, which is not a good habit. Some institutions have decided that duplicate therapy alerts should be turned off for providers and that the pharmacists should be the first to process these according to set guidelines, similar to the workflow in place pre-CPOE. Certain system setups have provided a helpful way to assist in combating this issue by allowing the parameters for number of duplicate warnings by class that can occur to be set. For example, opioid warnings or alerts could be set to occur when the number is only three or greater. This again is reminiscent to the pre-CPOE workflow, as when the pharmacist performed this screening process, many times they would only call a provider if two of the same or similar drugs were being used to treat the same level of pain.
When it comes to warnings for dose ranges on drugs, they are by nature complex due to the simple fact that many drugs have different high dose and low dose warnings depending on route or use. An example being, antibiotic dosing which can be different if given as a continuous method or intermittent. If the mapping of the dose ranges are not correctly set up, then a provider can get flagged or alerted for every dose that the medication is given regardless of the route or use. Clearly, this scenario has the potential to contribute to a provider’s alert fatigue.
Electronic clinical decision support in the form of alerts has not come to the point yet where it can replace the filters that an educated person can apply to a scenario. The only way to combat this fact, is for a stakeholder committee to regularly and continuously monitor and review the information surrounding overrides and missed alerts. As the flagging of these instances becomes more consistent and the pool of data to be analyzed grows, it can then be acted on accordingly with the hope of preventing future similar occurrences. After all even IBM Watson uses continual input of data to improve the capabilities to diagnose.
About The Author: Bob Beagley is a Pharmacist that currently serves as a pharmacy professional at HealthNET Systems Consulting, Inc. He has 30 plus years of experience in Healthcare IT and hospital pharmacy management including pharmacy implementations, system upgrades, process redesign, workflow/system build assessments and end user training.