Technologies designed to increase patient safety and operational efficiency are continually introduced into the health care environment. Although the safety benefits that such tools afford patients cannot be disputed, the operational and time commitments required for proper utilization can be significant. These demands, coupled with the increasing complexity of health care technology, can create significant opportunities for error.
This confluence of opportunity and risk is evident with smart infusion pumps. Smart infusion pumps offer robust safety and efficiency benefits, but their utility is dependent on their integration with the electronic health record (EHR) and the availability of a comprehensive drug library. Without proper training and experience, clinicians implementing smart pumps may fail to ensure that they are operated as intended and with their safety systems fully employed. Integration presents a unique opportunity to minimize human interaction (often the primary source of error) and thus maximize smart pump benefits. In this endeavor, dedicating sufficient time and resources prior to pump go-live is vital to minimize rework and to ensure the pumps deliver the greatest benefit.
Initiation of the Smart Infusion Pump Project
WellSpan Health is an integrated health care delivery network located in South Central Pennsylvania with eight hospitals, more than 170 outpatient patient care locations, 1500+ employed physicians, and over 19,000 employees. The health system’s pharmacy is a leader in the deployment of pharmacy automation and was one of the first hospitals in the state to implement robotic pharmacy dispensing technology. Since that time, the pharmacies have shifted to a decentralized dispensing methodology supported by carousels in the pharmacies and automated dispensing cabinets at the point of care. Other technologies employed across WellSpan Health include CPOE, BCMA, automated parenteral nutrition compounding technologies, clinical surveillance software, and Web-based clinical decision support optimization tools.
WellSpan Health completed a capital budget evaluation of smart infusion device technology in 2008, when our current infusion devices were reaching the end-of-life phase and the manufacturer indicated they would no longer be supported, because the lack of smart pump functionality made them obsolete. Although the pumps were near the end of their usefulness, a decision was made to wait a year prior to making this significant capital investment in order to take advantage of vendors’ ongoing smart pump development.
One year later, we were approached with an opportunity to enter into a development partnership with both our EHR vendor and an infusion pump vendor to participate in an integration project. By that time, the vendor’s pump development had been completed, which allowed communication of the infusion medication order from the EHR to the infusion device—ie, unidirectional communication. This represented a substantial step forward in improving the safe administration of infused medications at WellSpan. The vendors were now seeking to complete the integration of the infusion device back to the EHR, thus developing bidirectional communication, which would ensure the accuracy of nursing documentation for infusion administration and facilitate our exploration of additional opportunities for improving nursing and pharmacy workflows.
Achieving a bidirectional flow of information between the EHR and the pumps is key to preventing human error and maximizing pump benefits. Once infusion administration is initiated, the device communicates information back to the EHR automatically, ensuring proper documentation. Although developing a bidirectional interface between the EHR and smart pump is complex and can be time-consuming, the safety and efficiency dividends are significant. Our need for new devices, coupled with the opportunity to be intimately involved in the development of this important functionality, was an excellent opportunity. Thus, we agreed to participate in the partnership and began a pilot program in a 10-bed medical/surgical intensive care unit in WellSpan Health’s largest hospital.
The decision to simultaneously implement smart infusion device technology and bidirectional EHR integration offered a substantial benefit to our organization versus implementing smart pumps first and then creating a bidirectional interface with the EHR thereafter. Building a drug library to support pump integration requires a significant amount of consideration versus the more straightforward formulary review and product build within the EHR as required for a typical smart pump implementation. Key to this build was ensuring exact alignment between the formulary, EHR, and drug library. Completing this work simultaneously with the initial smart infusion pump library build meant that we did not need to backtrack and rethink an existing build, which could have proven significantly more time-consuming. Understanding the requirements for alignment at the initiation of the project streamlined the pump library build, which was advantageous considering WellSpan Health’s aggressive timeline of 3.5 months for completion of the pilot project.
One potential challenge involved the lack of a BCMA solution at WellSpan Health at that time (BCMA has since been implemented). This was a concern, as infusion pump integration relies heavily on bar code linking of the patient, pump, and EHR. At the time of the project, we had only implemented positive patient identification by bar code scanning; therefore, the new workflows created by infusion device integration represented a potential challenge to the project, as nurses had little experience in scanning medications at the bedside. Fortunately, pharmacy had extensive experience surrounding bar code scanning through the use of various bar code-dependent technologies within the pharmacy, such as robotic dispensing, and was able to educate and assist nursing in this regard. With pharmacy’s help, nursing quickly became competent at bar code scanning.
It became clear early on that infusion device integration would require the efforts of an integrated team to achieve success. Including pharmacy, nursing, IT (both technical and informatics resources), and biomedical engineering was critical. In fact, the large number of team members required project management resources to maintain the timeline and to ensure that all roles were present and engaged. Over the life cycle of the project, the role of the biomedical engineering team changed. Initially, they were only responsible for the day-to-day maintenance and distribution of the infusion devices across the facility; subsequently, they assumed responsibility for creating wireless interfaces to facilitate connectivity with the EHR. As a result, a new position was created, Medical Equipment Integration Specialist, to support the endeavor.
Developing the Drug Library
Compiling the smart pump drug library was a particularly critical element of the project. An early challenge was our lack of in-house experience with smart infusion pump technology. Prior to integrating the smart pumps with the EHR, we wanted the nurses to gain experience using the pumps with the dose error reduction software (DERS). Therefore, developing the drug library efficiently so the nurses could use it was paramount. Our goal was to minimize the number of drug library changes that would be necessary after the EHR integration, as we felt numerous changes could confuse end users. In this effort, standardizing our process and soliciting nursing input were important factors to success.
Pharmacy informatics staff members responsible for the EHR build partnered with pharmacy project staff to ensure alignment with the pump when integrated. Standardizing how drug volume would be accounted for in the final total volume of the infusion was critical. Considerations for pharmacy admixture processes were taken into account, and the build was adjusted to ensure that the library entries for infusions exactly matched the EHR build. Continuous infusions, which are frequently concentrated, require library entries for each concentration to ensure a match. Therefore, we adopted a standardized policy for concentrating these drips. Another prominent consideration was how to handle occasional infusions for non-formulary medications. In the interest of supporting consistent use of the drug library, we decided to review all infusion history over the prior 12 months and include all infusions administered during that period, regardless of current formulary status.
The importance of undertaking a robust build cannot be overstated. Because integration between the EHR and smart pump library is only successful when the library entries match exactly, it is necessary to build out more infusions, at a greater level of specificity, than previously necessary. This approach has the added benefit of preventing users from opting out of using the library and accessing generic infusion functionality, which bypasses the safety benefits of both the integration and the DERS software. Medications that can be administered as different infusion types (ie, both intermittent and continuous infusions) require special consideration, as each infusion must have a unique identification number that matches the infusion to the EHR formulary.
Input from Nursing
After the majority of the library was built, collaboration with nursing leadership and end users was necessary to determine how the library would present in different care areas to facilitate ready administration, and also protect against inadvertent inclusion of medications in areas where they are not appropriate (eg, vasoactive infusions in non-telemetry care areas).
Following the basic library build, multiple nursing end users, representing each of the care areas, were engaged to review the build to ensure that the medication concentrations and distribution of medications among the different care areas/profiles were appropriate. Pharmacy provided these representatives with a printed copy of the library representing their particular care area, as well as an electronic device with the library loaded, which allowed them to test the limits of various infusions to ensure they met expectations for the given practice. Data were collected to identify opportunities for improvement, and changes were made to the library based on user comments.
With the library sufficiently developed, it was then deployed to the new infusion devices on an interim basis without EHR integration, simply to allow nurses to develop familiarity with the technology. Once the devices were deployed across the organization, we moved forward into the integration testing phase to begin validating that the library build would support the auto-programming of the infusion devices from the EHR. Initial testing involved ensuring that we could establish a connection between the EHR and the infusion devices for the three basic infusion types: continuous infusion, intermittent infusion, and IV fluid infusion. This testing, which involved a small subset of the drug library, is an important first step toward later testing 100% of the library entries.
The testing phase required a considerable amount of time and dedicated resources. Due to the specificity of every infusion—for example, each had a specific concentration, volume, and alias/identification number—we found it imperative to test every single line item in the library. This was an extensive undertaking considering the large size of the infusion library; however, this level of detail is crucial to success. When end-users experience only minimal to moderate success using a newly implemented technology, they will quickly look for work-arounds rather than continue to navigate the poorly functioning technology. With this is mind, thorough testing is imperative to ensure all integration issues are alleviated prior to implementation. Approximately 8-10 individuals spent about 32 hours total on testing in their areas; although this required a significant commitment, it was time well spent.
Nursing education was initiated during the testing phase, prior to the launch of the pilot project. Nursing super-users were trained during the testing phase with the expectation that they would become familiar with the library functionality and then help all nurses gain a comprehensive understanding of how the pumps worked and be able to answer questions from their fellow nurses in the pilot area. This approach was particularly effective, as the super-users witnessed both the problems that arose and the solutions that were developed in response, giving them a strong understanding of the pumps’ capabilities.
Implementing Smart Pumps
After testing was completed, the pumps were implemented in the pilot area. Classroom education for the nursing end users proved necessary to thoroughly explain the functionality and demonstrate proper use. Although the new pumps ultimately simplified the nursing workflow by replacing numerous pump data entry requirements with a few bar code scans, classroom education underscored the value of integration to nurses and helped gain their buy-in for the change. For example, education was critical to helping staff understand the importance of reporting errors when the pumps did not work as expected, rather than simply moving on and leaving the next user to encounter the same issue.
Because this was an alpha project with the vendor at the time of our implementation, a small pilot in a single unit was the only reasonable way to implement the technology. Taking a deliberate, stepwise approach enabled a smooth implementation, as it allowed the project team to effectively hone in on issues and concerns.
The presence and availability of the project team was essential to the success of the project. We found that many of the issues encountered early in the implementation required input from various team members to identify a resolution. Ensuring the entire project team is available throughout the implementation phase is highly recommended when going live with smart pumps.
Setting Appropriate Alert Limits
Smart infusion pumps are intended to prevent medication errors by alerting users when doses exceed certain thresholds. However, a large number of clinically insignificant alerts must be avoided, as they create the potential for alert fatigue and reduce nursing efficiency. Thus, active management of smart pumps alerts is vital.
Any time an infusion violates a soft or hard limit, additional steps toward intervention are required on the device and in the EHR, which decreases the value of the integration in improving efficiency. Early in our post-implementation phase, we discovered that some of our overly restrictive upper soft limit dose alerts were causing these inefficiencies, and thus, a measure of dissatisfaction for nurses. As a result, significant effort was put into reviewing the infusion data reports from the devices to determine when and where the limits may have been overly restrictive so that adjustments could be made to reduce the number of alerts. With numerous formulary changes and various prescribing habits across our organization, we have found it important to continue to monitor this data on a quarterly basis (after initially monitoring monthly) in order to sustain efficiency over the long-term.
Results of the Pilot Study
While safety was the primary motivator for implementing infusion device integration, the resulting workflow efficiency for nursing end users was also extremely important. Manual programming previously required an average of 20 pump manipulations to prepare an infusion for administration; infusion integration reduced that number to fewer than five, and the majority of those steps are easily accomplished via bar code scans. Following successful integration, nursing realized further benefits on the back end of infusion documentation: the manual entry of values from the pump into the nursing flowsheets was replaced by automated data entry that only requires nursing to verify the data. The end result of this efficiency was a greater than 50% time reduction for nursing surrounding the infusion administration/documentation process.
The quality of nursing infusion documentation improved dramatically as a result of the smart pump implementation. The infusion devices capture a significant amount of data about the infusion even when not integrated; now this information can readily be transferred to the EHR. As a result, the accuracy of infusion documentation surrounding time-critical situations, such as a patient code, has dramatically increased compared with nurses’ former techniques of writing this information on paper towels, scrub pants, and even directly on their arms.
Our pump vendor supplies reports that gauge user compliance with auto-programming as an indicator of the success of the implementation. Secondary indicators of a success can be gauged via the number of drug library updates necessary as well as the number of infusion-related events that occur throughout and following the implementation. At our organization, the drug library is updated quarterly.
Rolling Out Smart Pumps Across the Organization
Subsequent to our initial smart pumps pilot, we applied a stepwise approach to guide the expansion of our integrated pumps across the organization to additional patient care areas and ICUs. Initially the smart pumps were present in every nursing unit, and were used without integration. Once the pilot was completed, we rolled out our stepwise implementation one unit at a time. This approach facilitated a focused implementation in each area and ultimately, a series of successful implementations with a high rate of compliance using the smart pump functionality.
Unexpected Device Switch
Approximately 5 years into our journey utilizing integrated smart pump technology, the unexpected retirement of our infusion device platform forced the organization to re-evaluate the device market and select a new smart pump. Thus, in 2015 we implemented new infusion devices and integrated the pumps with the EHR yet again. When we switched vendors, a new library build was required. Although some parameters of the library build were different under the new vendor, we were able to implement all of our existing integrated areas at one time, as the nursing workflow for programming the devices was largely unchanged due to the EHR-centric nature of the integration workflow. Because the majority of the workflow focused on the EHR and not the pumps, the change in devices was largely a non-event for the nurses. This experience highlights the importance of proper planning and a careful integration process in addressing unforeseen pump challenges moving forward. Dedicating adequate resources and attention to the EHR at the initiation of the smart pump project ensured we had the bandwidth to overcome this additional, unexpected challenge.
Moving to a Single EHR
Eighteen months after the 2015 implementation of the new vendor’s infusion devices, nurses appeared comfortable using the new pumps, and the project was largely in maintenance mode. At that time, due to rapid health system growth that resulted in the simultaneous deployment of three different EHRs across WellSpan’s five inpatient hospitals, the organization made the sensible decision to implement a single EHR application across the entire organization.
Fortunately, the EHR vendor that was ultimately selected also had infusion device integration experience; however, they had never implemented integrated infusion devices enterprise-wide. The vendor advised us that we would need to temporarily halt infusion device integration until such time that we had deployed the EHR successfully across the health system. Our organizational leadership, having firsthand knowledge and experience with integration through the pilot project, rejected this proposal, and we moved forward with a plan to implement the new EHR and smart pumps simultaneously.
The project team reconvened into active implementation mode. Pharmacy standardization and formulary changes required building yet another pump library (ie, our third experience rebuilding a library). Beyond the library build, we needed to identify the required workflow changes when using the legacy system versus the new EHR. The team found that while the build and set-up processes for the two EHRs were significantly different, the core functionality and workflow for end users did not represent a significant departure from what they were used to. As a result, outside of the educational efforts for all employees surrounding the use of the new EHR, integration-specific education was only required for nursing informatics staff members who traveled from unit to unit. The training was accomplished via unit-based staff meetings and by shadowing staff.
Looking to the Future
Our vast experience with integrated smart pump builds facilitated a smooth system-wide implementation of the new EHR and helped us identify opportunities for improvement in the 10 months we have been live. Despite our long journey with infusion device integration, additional growth opportunities exist for these devices, including enhanced functionality and further integration. The integration of additional types of pumps, such as epidural pumps and patient-controlled analgesia, will serve to further safeguard the processes surrounding the administration of essential medications. This integration is currently underway at WellSpan Health.
As smart pump technology matures, the safety and efficiency benefits will increase commensurately. WellSpan looks forward to keeping abreast of new technology advancements to ensure our patients and staff members continue to reap the rewards of smart pump use.
Donald “Chip” Gerhart Jr, RPh, is the manager of clinical pharmacy informatics at WellSpan Health in central Pennsylvania. He is a 1996 graduate of the University of Pittsburgh School of Pharmacy. Chip’s professional interests include the use of technology to support patient/medication safety, electronic clinical decision support tools, and interdisciplinary collaboration to improve patient care.