Readers Write: Medication Decision Support Alerts Don’t Need to Go Away, They Need to Get More Specific
Medication Decision Support Alerts Don’t Need to Go Away, They Need to Get More Specific
By Bob Katter
Bob Katter, MBA is president of First Databank (FDB) of South San Francisco, CA.
It’s no secret that clinicians are inundated daily with alarms and alerts that interrupt their workflows and cause cognitive overload, contributing to the industry-wide problem of clinician burnout. The National Academy of Medicine (NAM) even declared clinician burnout to be an “epidemic,” citing improved usability and relevance of health IT as one of six goals focused on addressing our current healthcare crisis.
While medication alerts are only a portion of a comprehensive clinical decision support (CDS) system, they contribute significantly to clinician alert fatigue. Clinicians are presented with an abundance of low-specificity and interruptive medication alerts and may even overlook critical alerts while sorting through the noise. This contributes to physician burnout and likely compromises patient safety.
We need to do better.
The good news is that given the wealth of patient information now available in electronic health record (EHR) systems, low-value and non-specific medication alerts can become a thing of the past. Medication alerts displayed to clinicians today can be patient- and workflow-specific, resulting in greater relevancy and efficiency.
Health systems and hospitals should focus on replacing non-specific medication alerts with more targeted alerts based on information from the patient’s chart, while delivering these alerts at the most actionable points in the clinical workflow. This approach helps reduce clinicians’ alert burden and fatigue, increases efficiency, and results in better clinical decisions and patient outcomes.
Origins of Alert Fatigue
Drug-allergy and drug-drug interaction alerts were among the first types of CDS alerts introduced in the heyday of EHR implementations. They were required as part of the Centers for Medicare & Medicaid Services (CMS) EHR Incentive Program, commonly known as Meaningful Use, and remain part of the mandatory functionality in 2015 Certified Electronic Health Record Technology (CEHRT). But they can be made better.
The number of data sources and the amount of healthcare information flowing into EHR systems has increased exponentially since the original introduction of these systems in the 2000s. With the level of patient-specific data, clinical guidelines, research findings and other critical information now available, EHRs can and should deliver more relevant and targeted medication information.
Here is how we could flip the script on medication decision support to create greater specificity, reduce alert fatigue, and ultimately improve patient safety and outcomes.
1. Make Alerts More Meaningful and Actionable
Decision support alerts that rely on medication lists alone are helpful but often limited in the insight they offer clinicians. We can create more relevant prescribing guidance by factoring in not only standard demographic information, but also other patient-specific context, including lab values, genetic test results, patient care setting, clinical risk scores, and comorbidities.
Due to advances in diagnostics technology and in IT systems interoperability, this information is more easily accessible than ever, creating opportunities to support more precise guidance and better outcomes. A deeper dive into patient information can help clinicians evaluate risks for complications such as hyperkalemia or QT prolongation. It can also help quantify patient risk for issues such as opioid addiction and a whole host of adverse drug events.
2. Consider the Scenario
Building context around medication alerts should also include the clinical scenario. When a patient has just undergone heart surgery, for example, standard care guidelines typically recommend administering multiple medications post-surgery that would not normally be taken together. Although some of these medications may interact, which could be problematic in another context, these interactions can be monitored and managed in an acute care setting. In this case, surfacing standard interaction alerts would not increase patient safety but would create unnecessary noise.
3. Build it in the Workflow
In another study of CDS usage, one of the obstacles to clinician adoption cited was “disruption to workflow,” a common complaint about medication alerts. When evaluating drug risks, clinicians may need to search through the EHR or log in to a lab results portal to verify the information and to ensure that the alert is relevant. This slows them down and distracts from patient care.
Health systems should present relevant alerts with adequate supporting data when and where they are needed in the workflow. For example, when a patient’s potassium levels have reached a specific threshold due to an ongoing drug-drug combination therapy, the EHR should initiate an alert at the right point in the workflow when the issue can be best addressed.
This is not meant to say, however, that alerts presented at the point of ordering cannot be useful in some cases. For example, a general reminder to order a blood test to check potassium levels when ordering a certain drug therapy can be followed by a patient-specific alert later in the workflow to adjust the dosage once the lab results are returned.
4. Focus on Specificity
According to a recent study, clinicians are more likely to accept and act on CDS guidance when presented with patient-specific alerts based on EHR data.
Reducing quantity and repetition of alerts is also important, considering a recent study of clinicians found the likelihood of alert acceptance dropped by 30% for each additional reminder received per encounter. Reducing generic alerts and improving the patient specificity of the remaining alerts would go a long way toward improving the acceptance rate.
5. Optimize the CDS
Health systems should continually analyze how their clinicians are interacting with alerts and whether the alerts are doing more to protect patient safety or to distract providers. By reviewing the data generated during the medication ordering process, health systems can predict how clinicians will respond to specific alerts and strive to generate only those alerts that help clinicians make better decisions and ultimately protect patient safety.
Putting Patients First
The bottom line is that medication alerts do not need to go away, they need to get more specific. By taking a deeper dive into the relevant information about a specific patient, at the appropriate point in the clinician’s workflow, decision support can deliver more meaningful and actionable insights. If such a patient-specific approach were to be deployed across the industry, we could significantly reduce the cognitive burden that these systems place on clinicians while simultaneously improving medication-related patient safety.