Smart pumps can substantially bolster medication safety initiatives, but for this technology to succeed, efforts cannot end at implementation. The success of smart pumps hinges on ongoing improvements to the drug libraries and practices, as well as close collaboration between pharmacy and nursing to optimize use. Drug libraries need to be user-friendly and accurately reflect practice standards; otherwise, the safeguards provided through smart pumps are rendered ineffective. The data available through many of these pumps can be a powerful tool in driving improvements that ensure effective drug libraries and adherence to them. Since implementing smart pumps in August 2008, we have been using the smart pump reports to improve compliance and create more meaningful alerts.
Analyzing Smart Pump Data
For the first nine months after implementing smart pumps throughout Sanford USD Medical Center, we aggressively reviewed the report data to optimize the drug libraries in place. Our goal was to look at overrides of soft alerts and edits to hard alerts so we could assess where there was a need for practice changes and identify where more meaningful alerts needed to be created. We collected and analyzed data on a unit-by-unit basis—pulling three to six months’ worth of data from each unit. We looked at edit variance detail, which provides information on how programming was changed due to an alert, and override variance detail, which shows the frequency and percentage a specific medication was run outside of the predetermined soft limits. We also analyzed the most commonly used medications on each unit over a three-month period in order to identify the need to change the top six medications, which are those medications that appear on the smart pump screen when nurses initiate programming. After reviewing the data, we shared our findings with frontline staff and leadership on the specific units, and looked further into work practices, daily interactions with the pumps, and possible barriers to efficient pump programming in order to pinpoint the necessary changes that would encourage fewer edits and overrides. By working closely with nursing staff, we were able to make specific, informed changes to drug libraries and, when necessary, to the IV medications prepared in the pharmacy. We also were able to identify the need for educational opportunities for nursing staff in cases where alerts were appropriate, but were not acknowledged as such.
Another constructive report is the pump event log, which helps in investigations of safety issues regarding pump function. This report provides a dated and timed line-by-line review of every key stroke, action taken, and alarm on the pump, whether automatic or initiated by an end user. In addition, the event log notes which clinical care area was used for the initial entry. More often than not, this report provides evidence of user error rather than pump failure. The event log is a useful tool for walking staff through specific situations in order to explain the sequence of pump programming and alarms. In one instance, we used the event log to help uncover the reason for a heparin infusion delay that was initially blamed on pump failure. The nurses reported that the pump was checked and infusing at the beginning of the night shift; however, when morning labs were drawn, bleeding times were substantially lower than expected, and the pump was found unplugged and off. After reviewing the event log, we discovered that the pump was in fact not running as reported and appropriate alarms had sounded indicating the pump was shutting down due to inadequate power. Using the event log, we were able to show the nurses that the pump had already shut down by shift change and that alarms, which were most likely overridden, had gone off prior to the pump shutdown.
We eventually plan to use bar coding technology with the pumps to select the appropriate drug and concentration for administration to the patient and subsequently document it within the electronic medication administration record. With this ability, the event log will provide even more detailed information by connecting all pump data to a specific patient and end-user.
Following the initial work with each clinical care area, we turned our focus to managing medications that are included in the ISMP high-risk medication list. In addition, we explored how to best use the executive scorecard (see Figure 1); this report was developed by our vendor to provide an overview of compliance by highlighting the top 10 medications with the most alerts, the top five infusions that had their alerts overridden, and the top five infusions where an alert resulted in an edit to the program. This assists us in pinpointing areas of highest concern. The override data on these reports helps us assess if the drug library indeed reflects prescribing and nursing infusion practices, as well as whether the medication is being delivered based on the established library limits. The edit data on these reports provides evidence of the safety benefit of the smart pump technology. In addition, these reports create learning opportunities for staff. We have uncovered numerous situations where staff does not trust the set limits and continues to reprogram the pump in the same manner multiple times. This happened in one case with heparin, which was reprogrammed 17 times on one occasion. This reluctance to trust the established limits underscores the importance of sharing the data from these reports with staff and providing continual education as to how limits are determined. Unit meetings and medication memos are good vehicles for delivering this type of education.
Improving Soft and Hard Limits
Soft limits on a smart pump are tantamount to a tap on the shoulder; they are intended to encourage the pump operator to stop and consider whether the right dose/rate is being programmed. While these alerts encourage critical thinking, if encountered too often they can lead to alert fatigue. Using data from the pumps, we can reduce the potential for alert fatigue by uncovering soft limits that are firing multiple times—allowing us to investigate these occurrences and then implement the necessary changes, either in practice standards or the pump library. For example, when the pumps were first implemented, we discovered that antibiotic soft limits were overridden frequently. With the data available, this was traced back to the overfill amounts in the antibiotic IV bags—both pre-filled and compounded; limits in the drug libraries were adjusted and new practice standards for preparing antibiotic IV bags were employed in the pharmacy to accommodate the nursing workflow.
Smart pump data analysis has led to hard limit improvements as well. For example, we discovered that underdosing of heparin was a frequent issue, as nurses were overriding the established lower soft limits and no lower hard limits were in place. In one case, a patient was admitted with a bilateral pulmonary embolism. Heparin infusion (100 units/mL) was initiated on admit. The ordered dose was 1700 units/hr, but the nurse programmed 17 units in the pump thinking about the dose in terms of mL/hr instead of units/hr. The nurse received a lower soft alert but overrode it. This subtherapeutic infusion continued for over three hours before it was recognized. This variance forced us to evaluate how often this was occurring. Six months’ worth of smart pump data on heparin was retrieved, which pointed to a significant need for lower hard limits on heparin infusions to prevent subtherapeutic doses, which can be as detrimental as heparin overdoses (see Figure 2). Based on this information, a lower hard limit of 300 units/hr for heparin infusions was established. While we have made significant improvements to our drug library based on the pump data, we continue to explore opportunities to establish more upper hard limits as well as lower soft and hard limits.
Value of Collaboration
Collaboration between nursing and pharmacy was essential to the success of the drug library build and it continues to be critical to further pump optimization efforts. Management of the drug libraries is delegated to representatives from pharmacy (ie, two medication safety officers, who manage four drug libraries) in conjunction with the clinical nurse leader. Recommendations for changes to the drug library are sent to both the nursing and pharmacy representatives so both teams can weigh in. The clinical nurse leader investigates the nursing side of the issue, while the medication safety officers discuss IV preparation with the sterile products coordinator. Changes to the library are made collaboratively and then reviewed to ensure feasibility from both perspectives.
Smart pumps can play a crucial role in improving medication safety, but only if continued efforts are undertaken to optimize their use. Taking advantage of the data available through our pumps has allowed us to create more meaningful alerts, adjust practice standards when appropriate, and provide education to better facilitate understanding of the value these pumps bring to patient safety efforts; compliance with our drug library now stands close to 100%. Collaboration between nursing and pharmacy was and continues to be vital to our success with this technology.
Erica DeBoer, RN, is the clinical nurse leader at Sanford Medical Center in Sioux Falls, South Dakota. Erica graduated with her bachelor of science in nursing in 1997 from South Dakota State University. She has worked in a variety of acute care settings including pulmonary and medical surgical units, spending the majority of her time in the adult critical care setting. Erica completed her master of arts in nursing in 2010 from Augustana College. The emphasis of her current work involves the application of evidence-based practice guidelines, program development, patient safety, and developing hospital-wide processes that are efficient, data-driven, and focused on obtaining better patient outcomes.
Megan Maddox, PharmD, BCPS, is the medication safety officer at Sanford Medical Center in Sioux Falls, South Dakota. She received her PharmD in 1998 from South Dakota State University, followed by a specialty residency in family medicine at St John’s Mercy Medical Center in St. Louis, Missouri. After completing her residency, Megan held faculty positions at St. Louis College of Pharmacy in Missouri and Campbell University School of Pharmacy in North Carolina, where she also served as a clinical pharmacist in Duke University’s family medicine residency program. Prior to her current position, Megan was employed with GlaxoSmithKline as a medical information scientist and then a regional medical scientist supporting their cardiovascular, respiratory, and endocrinology products. She is a board-certified pharmacotherapy specialist as well as a certified diabetic educator.
Smoothing Out the Kinks with Smart Pumps
Despite careful planning, we still encountered many challenges during the initial transition to smart pumps. When the pumps first went live on the units, nurses were deafened by constant proximal occlusion alarms and air-in-line alarms. The proximal occlusion alarms were caused by tubing bending over the top of the cassette, which is easily resolved by placing tubing in the tubing guide; however, nurses found the alarms annoying and often did not take the time to read the alert on the pump to assist with problem solving. The air-in-line alarms were caused by bubbles collecting in the cassette. While this does not cause patient harm, the bubbles are still detected by the sensor that prevents air delivery to the patient. We resolved this issue by employing new tubing with a microbore segment that contained the bubbles.
We experienced additional tubing challenges with back-check valves not working as intended, which allowed medication from the piggyback bag to travel up into the primary infusion bag. This was first noted when a chemotherapy agent (which is red in color) was being infused continuously based on specific dosing, and it crept up into the primary bag, rendering the flush bag and the therapy unusable. Representatives from different departments came together to identify a more safe and reliable solution. We made the decision to try blood tubing, adjusted our standard method for chemotherapy delivery slightly, and designed a new set of tubing called the ball-in-chamber tubing. This new tubing configuration has worked well for us since its implementation.
We also dealt with some software challenges after first employing the pumps. One very concerning software issue surfaced after a software upgrade. Not long after the update, we received reports of pump screens turning white and freezing, forcing the pumps to shut down and restart. After much discussion with multiple staff members, we isolated the steps that created this software error, and an additional software update remedied the problem quickly after it was identified.
After rollout, extensive support from our vendor, as well as constant reporting of issues and continuous follow-up, helped facilitate rapid resolutions and nurse buy-in. Maintaining open communication and sharing updates and education tidbits were part of our weekly workload for the first several months to ensure that continued optimization of this new technology occurred.
Tips for Creating a Meaningful Library