Providing patients with safe and effective anticoagulation therapy is a challenge facing every hospital, as the serious risks of thrombosis and hemorrhage leave little room for treatment error. Yet compliance with TJC’s National Patient Safety Goal for anticoagulation has proven to be difficult for many hospitals. Appropriate monitoring of anticoagulation provides an opportunity for pharmacists to improve direct patient care and ensure compliance. Follow-up monitoring of anticoagulation is critical to improve patient outcomes.
Anticoagulation management teams, often pharmacist-based, are an excellent way to ensure this goal is achieved. Unfortunately, creating these teams is not always feasible due to resource limitations. At Vanderbilt University Medical Center (VUMC), clinical dashboards are used to monitor anticoagulation therapy. These dashboards are online sites that list all patients receiving the focus medications. Our facility currently monitors warfarin, unfractionated heparin, and enoxaparin using dashboards.
At VUMC, dashboards were developed to facilitate the accurate monitoring of high-risk medications and promote compliance with hospital anticoagulation protocols and policies. Our primary focus was to prevent adverse drug events caused by supratherapeutic anticoagulation.
There can be more than 100 patients receiving therapeutic anticoagulation each day at VUMC; with this large number, it is impractical for the pharmacy to monitor every patient. By utilizing dashboards, pharmacy can narrow the focus to the 10-15 patients per day who must be evaluated for safety. The creation of specific alert values allows pharmacists to identify and focus on those patients who are most likely to be at risk of adverse events, and significantly reduces the total number of patients that need to be reviewed. The pharmacy is now able to monitor patients throughout the entire medical center without any extra FTE pharmacy resources, thereby improving cost savings.
Developing Appropriate Alert Values
Quality informatics support is key to successful dashboard development. Clinical dashboards are most successful in health systems with electronic medication administration records (eMARs), as the pharmacy and laboratory systems must be able to communicate with each other seamlessly before any alerts can be created. The main goal when developing and setting alert values is that they are appropriately sensitive⎯—identifying and monitoring patients at risk of supratherapeutic anticoagulation⎯—⎯without classifying patients as at risk who are not, thereby producing too many false positives. A large number of false alarms can lead to alert fatigue, and pharmacists may begin to ignore the warnings. To ensure appropriate alert values, they should be developed by the pharmacists and other clinicians skilled in anticoagulation management who will be responsible for responding to them.
Alert values are developed based on an expected outcome for a target medication, such as how rapidly an International Normalized Ratio (INR) should rise on warfarin. We have also developed alerts based on monitoring requirements, per hospital policy. Alert values need to be balanced by the possible output from an electronic system. For example, policy may require that an INR be checked daily, but an alert value for INR >24 hours old could produce far too many alerts to be a helpful parameter. An INR review performed after 25 hours due to a delayed lab draw is not going to cause patient harm. Extending the alert value to 36 or 48 hours is much more practical.
After our initial dashboard development, several modifications to the alert values were required to ensure accurate capture of patients who required monitoring, without creating false positives. For instance, our warfarin dashboard was configured to provide alerts for a rapid rise in INR (>0.4 in 24 hours), high INR (>3 in 24 hours), and old INR (no INR in the last 48 hours). Unfractionated heparin and enoxaparin are combined into one dashboard with alert values for old partial thromboplastin time (>6 hours), creatinine clearance (CrCl) <30 on enoxaparin, heparin infusion >2,500 units/hour, and platelet drop >50% for possible heparin-induced thrombocytopenia (see Table 1). We found that these values captured the most appropriate information to suit our needs. However, continual monitoring of the capture numbers is necessary to ensure optimal functionality.
Alert values must be constantly reviewed to identify better ways to assess patient risks and prevent adverse drug events. Determining what the alert value should measure and delivering the expected clinical results is not always a straightforward assignment. Our heparin dashboard required several alert revisions in order to deliver maximum functionality. The inclusion of prophylactic doses along with therapeutic dosing of anticoagulation created challenges due to the number of alerts. For example, the alert for patients on heparin or enoxaparin without a platelet count in the previous 24 hours created far too many alerts to be useful. To avoid alert fatigue, the current version of the dashboard only alerts for patients on prophylaxis if they have a CrCl less than 30 mL/minute on enoxaparin or have a platelet drop >50% on any heparin or enoxaparin.
Although the use of dashboards has been valuable in our facility from a safety perspective, the use of this technology must be accompanied by skilled pharmacist intervention. The technology alone is only as good as the trained staff who oversee it.
Our dashboards are designed to pull up a list of all patients at VUMC on each anticoagulant (Figure 1). Results are sorted so that patients with an alert value are listed at the top of the dashboard, calling the pharmacist’s attention to these patients. Clicking on the individual patient name opens a window that displays that patient’s critical information, such as lab values, anticoagulant drug orders, administrations via the eMAR, and any interacting drugs (Figure 2). The comment field enables the pharmacist to communicate interventions and a daily care plan.
Click here to view a larger version of this Table
Each day a single pharmacist is assigned primary responsibility for monitoring the dashboards, which are tracked seven days a week. The pharmacists monitoring the dashboards include clinical pharmacists, staff pharmacists, and pharmacy residents. Each clinical pharmacist reviews alerts for his or her specific area of clinical coverage, and the primary anticoagulation dashboard pharmacist is responsible for all other alerts for the hospital. Each alert is reviewed individually to determine if an intervention is needed.
When reviewing a dashboard alert, there is often enough basic information to make a rapid decision about whether or not an intervention is warranted. In many cases the primary team has already responded to the issue. The pharmacist then determines if the issue was appropriately addressed, or if further modification of the dose is needed. If the primary team has not addressed the issue in an appropriate manner, the monitoring pharmacist is responsible for further investigation. For the warfarin dashboard, it is important to evaluate the patient’s current clinical situation, the initiation or discontinuation of interacting medications, INR trends over time, and warfarin dosing history. For the heparin dashboard, the most frequent pharmacist intervention is for patients that are on enoxaparin with a CrCl less than 30 mL/minute. This alert is most likely to occur when a patient has had acute renal failure after enoxaparin is initiated.
The necessary interventions are often significant. Many times interventions result from an oversight by the provider, noncompliance with hospital protocols, or a lack of knowledge about anticoagulant therapy. For example, interventions have been triggered by the warfarin dashboard in situations where the primary team was set to discharge a patient with a supratherapeutic INR without reducing the dose, a patient with a supratherapeutic INR upon admission had their home dose continued, and a patient did not have an order for follow-up INR monitoring.
When these issues are uncovered, the prescriber is usually grateful that pharmacy has helped remedy a potentially dangerous situation. Once the alert is resolved, the pharmacist checks the appropriate box at the top of the screen to signify that the alert has been addressed for that day. The primary dashboard pharmacist is responsible for a final daily review to ensure that all alerts have been addressed. A pharmacy resident also is assigned this duty on a daily basis to perform a double-check.
To demonstrate the value of any project, outcomes must be evaluated. At VUMC, we track the rate of INR values greater than 4 to measure our success with the warfarin dashboard. Each week an email with the rate of INR values of 4 or greater, coupled with the patients’ medical record numbers, is sent to key personnel monitoring the warfarin dashboard. The resultant retrospective review of outcomes is used to evaluate pharmacist performance on the dashboard and also is a mechanism to identify opportunities to adjust or identify new alerts. Our initial experience has suggested that our rate of INR values over 4 may have been reduced by as much as 30% by dashboard use; we hope to verify these results in the future through a formal review.
Not all medical centers have the informatics support available to implement the use of dashboards; individual institutions need to evaluate their ability to support this type of project. For hospitals without the resources to develop and maintain this type of system, there are commercially available clinical monitoring programs that provide similar monitoring processes for pharmacy.
When developing a dashboard system, it is important to identify key pharmacists that are knowledgeable in anticoagulation to interpret the dashboard alerts and notify prescribers. Dashboards provide an excellent opportunity for clinical staff development, but skilled pharmacists are critical to successful dashboard implementation. The dashboard pharmacist should be able to interpret the alerts before notifying prescribers; each alert should not generate a call to the prescriber unless it represents a significant issue to the patient. Prescribers are less likely to accept recommendations if a large number of inappropriate phone calls are generated by dashboard alerts.
Keep in mind that it is not possible for dashboard alerts to prevent all adverse drug reactions—⎯patients have unanticipated jumps in INR that dashboards cannot predict. A skilled, human interface is necessary to interpret the alerts and recommend new dosing strategies. Ensuring consistent responses to alerts among the different pharmacists who staff the dashboard also can be challenging. To reconcile these potentially different responses, assign primary responsibility to a single staff member who oversees all dashboard activities.
Look for potential risk factors that can develop as a result of focusing on a single therapeutic goal. In the case of our warfarin dashboard, focusing on the prevention of supratherapeutic anticoagulation does create a risk for inadvertently promoting subtherapeutic anticoagulation. To prevent the systematic underdosing of warfarin, pharmacists follow up each intervention the next day to determine the resultant outcome and ensure the warfarin dose was not reduced excessively.
Our anticoagulation dashboards undergo constant revision to better predict patients that may be at risk of adverse drug events. Our goal is to find better ways to alert clinicians about drug interactions, integrate dietary considerations, and include potential disease state issues. Drug interactions are currently included in our dashboard, but there are no alert values generated as a result of these interactions.
In addition, our current system does not provide any alerts for patients with subtherapeutic INR values that may not be progressing appropriately toward their therapeutic goal. An alert for patients whose INR is rising too slowly, too rapidly, and one for patients with higher than average INR goals also would be beneficial. Our current alert for high INR fires for any value over 3, but many patients have INR ranges in which values over 3 may still be in the therapeutic range. A common example of this would be a mechanical mitral valve patient, for whom the INR goal is from 2.5 to 3.5. There are also many other patients with severe thrombotic states that require high INR goals.
New anticoagulant medications, such as dabigatran, also should be monitored via the dashboards; we are currently working to implement dabigatran alert values to flag patients with renal dysfunction. Other dabigatran alerts that could be useful include ensuring appropriate bridging and avoiding unnecessary monitoring. The successful use of dashboards at VUMC to monitor anticoagulation has sparked interest in developing a similar process to promote antimicrobial stewardship, monitor high cost medications, and ensure formulary compliance.
Our dashboards have proven to be a valuable tool in dealing with the challenges of managing anticoagulation therapy. Although the development of these dashboards took a great deal of effort, the return on investment has been significant. Our patients are safer and our pharmacists play a new, expanded role in direct patient care across a wide range of clinical services.
Daniel C. Johnson, PharmD, BCPS, is a clinical pharmacist specializing in cardiology and cardiac surgery at Vanderbilt University Medical Center in Nashville, Tennessee. He received his doctor of pharmacy from the University of Tennessee Health Science Center in Memphis and completed a pharmacotherapy specialty residency at Barnes-Jewish Hospital in St. Louis. Daniel’s clinical areas of interest include anticoagulation therapy, antiplatelet therapy, and ICU delirium.