Pamela J. Pockras, MSN, NNP-BC, CLC
& Reanna M. Smith, BSN, BA, RNC-NIC
Pockras, P. & Smith, R. (2013). Reconciling Pyxis overrides after the implementation of EPIC. Online Journal of Nursing Informatics (OJNI), vol. 17 (3), Available at http://ojni.org/issues/?p=2868
Cincinnati Children’s Hospital Medical Center (CCHMC) uses Electronic Privacy Information Center (EPIC) as the electronic medication record (EMR). The EMR is interfaced with the Pyxis or Automated Dispensing Machine (ADM), a decentralized, automated medication dispensing system located on inpatient units. When a medication is removed from the Pyxis without an active order (overridden in the Pyxis), the patient’s Electronic Medication Administration Record (eMAR) is automatically updated to include this medication. The focus of this article is electronic systems and how electronic systems interface with each other and how humans interact with electronic systems. Principles of improvement were used to identify the need for a practice change and the steps taken to implement that change. The goal of the practice change was to increase the percentage of medication reconciliation for those medications removed from Pyxis on override from the current 33% to 90%. Assuring medications are reconciled promotes patient safety and meets the requirements of the Center for Medicare & Medicaid Services (CMS) and the recommendations of Joint Commission Standard MM.8.10.
Key Words: EPIC, EMR, eMAR, Pyxis, overrides, ADM
Cincinnati Children’s Hospital Medical Center (CCHMC) started using the Electronic Privacy Information Center (EPIC) for all inpatient charting in January 2010. Before EPIC, emergency medications were overridden and removed from the Pyxis following a verbal order. Medication administration was documented in the electronic medical record (EMR) after an order was written by the ordering practitioner. This could be a delay of up to one to two hours after medication administration. This practice poses safety risks and was questioned.
Medications that are removed by override bypass many safety checks built into the system. First, overridden medications from the Pyxis have not been reviewed by a pharmacist. And these medications have not been validated using CCHMC’s multi-system check designed to help prevent medication errors. Additionally, nurses were receiving verbal orders from practitioners and omitting two important steps: entering the orders in the patient chart and conducting read back verification to the ordering practitioner. Nurses waited for written orders by the practitioner before charting administration of a medication removed on override from the Pyxis. This practice can cause medication errors, including a medication given to the wrong patient, incorrect dose, incorrect medication, or incorrect route.
After the institution of EPIC, the Pyxis and EPIC were interfaced. Medications removed via override from the Pyxis now automatically populate the Electronic Medication Administration Record (eMAR). By design, EPIC requires a medication override to be linked to an order when charting administration, thus reconciling the overridden medication. If the overridden medication is not reconciled with an order, the nurse cannot document medication administration in the eMAR. This results in a record of the Pyxis dispensing the medication without a record in the eMAR of administration. Without an order and eMAR documentation, the legal record reflects a nurse practicing outside his/her scope. These practice concerns were brought to the forefront due to the high risk of compromising patient safety and to comply with the requirements of the Center for Medicare & Medicaid Services (CMS). A multidisciplinary team was established to find a solution.
An initial review of the literature utilized the key search words EPIC, eMAR, Pyxis, overrides, and Automated Dispensing Machines (ADM). The initial search resulted in minimal information. Several months later, the search was repeated utilizing the same set of key search words: EPIC, EMR, eMAR, Pyxis, overrides, and Automated Dispensing Machine (ADM) and yielded the same minimal information.
In 2004, The Joint Commission (TJC) developed new standards for medication management. Standard MM.4.10 states “all prescriptions or medication orders must be reviewed for appropriateness” (Rich, 2004, p. 1353). The review includes details such as the appropriateness of the drug, dose, frequency, and route of administration (Rich, 2004). The exception to this standard involves situations where a licensed independent practitioner controls the ordering, preparation, and administration of the medication and urgent situations when a delay would harm the patient (Rich, 2004).
After TJC Standard MM.4.10 was established, hospitals reviewed processes for prescribing medications, utilizing pharmacy to double check the medication, and delivering medications. Pyxis machines were introduced to CCHMC before TJC’s developed new standards in 2004. The machines “improve the accuracy of pharmacy inventory and billing, streamline the distribution process, improve first-dose turn-around times, and aid in securing narcotics” (Paparella, 2006, p. 71). To improve safety, each hospital established a limited list of medications readily available for unit staff to access for overrides. Having a select list of medications in the Pyxis that can be overridden allows for quick access during an emergency situation, when pain management is needed, or sedation for a bedside procedure (Kowiatek, 2006). In the event of an emergency, nurses have the option to override the patient’s medication profile in Pyxis and obtain a medication from the select list. Overriding the patient’s profile in Pyxis bypasses the pharmacy review prior to administration and may lead to medication errors.
Two quality improvement projects related to medication administration were noted during the literature review. Kowiatek (2006) “utilized a continuous quality improvement approach to restrict the number of types of medications available on override status from the organization’s automated dispensing device” (p. 310). The first step involved assessing the types and frequencies of medications removed from their Automated Dispensing Device (ADD) on override. Then an expert panel established criteria for override access and revised the medication override list. A tool was also developed to monitor override medications. The audit revealed opioids were the category of medications most frequently removed from the Pyxis via override. To reduce the number of overrides and limit medication errors, the availability of opioids and sedatives in the Pyxis were limited to those most commonly used (Kowiatek, 2006). Additionally, the audit revealed that medication administration was not always documented on the medication administration record (MAR). Medications were also removed on override despite being already available on the patient’s medication profile. A multidisciplinary educational plan was implemented and a new policy written. As a result of their efforts, Kowiatek reported a significant decrease in the opioid override rates in the hospital and on each unit for the time periods between August 2003 and December 2004 (2006, p 313).
More recently, Early, et al., (2011) used a systems approach to improve the process for utilizing Bar Code Medication Administration (BCMA). This project was initiated after a root-cause analysis of a near-miss event revealed the impact overrides have on medication safety. A multidisciplinary team including nursing, pharmacy, human resources, quality management, and technology services was formed. The team defined an override as “an error message displayed as a result of the system’s inability to read a bar code or the bar code reading wrong medication, dose, and/or route” (Early, et al., 2011, p. 158). The electronic message was given to the clinician administering the medication, however the message was bypassed and the medication administered regardless of the warning. The team identified rationales for overrides as equipment problems, internal process problems, human factors, and non-barcode issues. Nursing feedback indicated overrides were viewed as an everyday occurrence and often accepted as part of the workflow. Equipment issues addressed included updating software and placing new BCMA scanners throughout the institution. An override algorithm was developed addressing acceptable reasons for overriding a medication. A second signature was also required if the override reason did not meet one of the acceptable rationales. Mandatory education was developed, including regular updates on project outcomes and operational functions. A significant decrease in the number of overrides was seen throughout the institution.
Our team utilized a quality improvement process that has been used by many industries outside of health care. CCHMC adapted information from the book The Improvement Guide: A Practical Approach to Enhancing Organizational Performance by Langley, Moen, Nolan, et al. (2009) to develop a quality improvement process that supports evidenced-based practice and aligns with the hospital’s safety initiatives. Key principles of improvement specified by Langley, et al., are “(1) knowing why you need to improve; (2) having a way to get feedback to let you know if improvement is happening; (3) developing a change that you think will result in improvement; (4) testing a change before any attempts to implement; [and] (5) implementing the change” (Langley, et al., 2009, pp. 16–20). We adapted each of these principles to drive our improvement process.
The Pyxis override resolution rate in the Regional Center for Newborn Intensive Care Unit (RCNIC; a level III-C rated NICU with 59 licensed beds and averages 700 admissions per year) was 33% prior to implementation of the project. The need for change was identified. As this poses a high risk to patient safety and gives the perception that nurses are practicing outside their scope of practice.
A small multidisciplinary team was assembled to address the issue of reconciling overridden medication. The RCNIC was the pilot unit. The team was developed by the Project Administrator for EPIC and comprised bedside nursing staff, an Advanced Practice Nurse (APN), the quality outcome manager for the RCNIC, and the unit pharmacist.
The Project Administrator pulled medication override data from EPIC reports daily. The data for previous failure categories and frequencies were shared in a Pareto chart. This provided the group with insight into the largest areas for improvement. This data was used to develop a SMART Aim (outcome goal). The SMART Aim was to increase the Pyxis override resolution rate in the RCNIC from 33% to 90% in 6 months. The percentage of resolved override pulls was monitored (numerator: number of resolved override pulls; denominator: total number of override pulls). The SMART Aim supported our global aim: to increase the total Pyxis override resolution rate throughout the institution and improve patient safety.
Before developing changes designed to improve Pyxis override reconciliation rates, the current process must be assessed.
The Project Administrator observed the current state workflow of staff nurses related to removing medications on override and reconciling the medication orders. The Project Administrator also conducted observations to better understand the future state; the impact on the clinical setting of the communication between Pyxis and EPIC. Many lessons were learned in this process. For example, nurses wanted to reconcile overrides, but did not know the process for reconciliation. Nurses habitually removed medications without an order and then bypassed a reminder in EPIC chart. This occurred due to a lack of understanding of this functionality. Nurses did not fully understand what caused an override or the interaction between Pyxis and EPIC. Sensitivity to time was also present with emergency medications.
From observations and input of team members, a simplified Failure Modes and Effects Analysis (FMEA) were created. A FMEA is commonly “used by process and product designers to identify and address potential failures” (Langley, et al., 2009, p. 411). The current process for overriding medications was analyzed. The failure modes/barriers to the completion of the process for medication reconciliation were identified. Potential interventions were also identified to facilitate the completion of medication reconciliation. After looking at potential override reconciliation barriers, three areas were identified: a lack of awareness that a Pyxis override occurred, a lack of knowledge by nurses on the process for reconciling a medication override, and algorithms for ordering medications did not match current practice.
From the potential failures and ideas for mitigating those failures in the simplified FMEA, a Key Driver Diagram (KDD) was developed. A KDD is a way to organize and summarize theories and ideas throughout quality improvement projects (Langley, et al., 2009, p. 429). CCHMC has adapted a version developed by Nolan, et al. (2009). This diagram served as a living document throughout the project and was revised as new information was received. The key drivers included utilization of EPIC and Pyxis to ensure resolution of all overrides, knowledge of the Pyxis override resolution function, staff buy-in of Pyxis override functionality, and the appropriate placement of orders in EPIC.
Using all available data and workflow information, the group selected to look at overrides categorized by medication. Heparin flushes and rapid sequence intubation medications were noted to be the highest percentage of medication overrides. As a result, these two types of medications were assessed, as well as on the global assessment of medication overrides.
The PDSA (Plan, Do, Study, Act) Cycle—“as a framework for an efficient trial-and-learning methodology”—answers the three questions, “What is the goal, how to determine if the change is effective, and what changes result in improvement?” (Langley, et al., 2009, p. 97)
The lack of awareness was the first barrier addressed in September 2010. A checklist used at shift change was modified to include checking the eMAR for unresolved overridden medications. This change was trialed for four weeks. At sign-out, off-going nurses were asked to check the eMAR and resolve overrides that may have occurred during their shift. It was predicted that increasing the nurses’ awareness of Pyxis overrides would result in increased rate of medication reconciliation from the current 33% to 70%. However, no appreciable change was noted. The change in the shift checklist was not being used routinely by staff and, when overrides were noted in the eMAR, nurses were still unsure how to reconcile the override.
In an attempt to increase the awareness of Pyxis overrides, email notifications of an existing override and instructions to resolve the override were sent to each nurse. These notifications were sent weekly by the Quality Outcomes Manager in the RCNIC. The predicted outcome was that resolution rates would increase over the next two months to 80%. Feedback received from the email notification stated the instructions were not clear and more education was needed for nursing.
An educational handout including the step-by-step process to resolving Pyxis overrides was developed. The handout was given to five nurses to pilot during their shift. The project manager received feedback that the educational handout was too general, more detail was needed, and hands-on training would be helpful. This feedback emphasized that the education gap was more significant than previously predicted.
A step-by-step process on how an override is created and how it can be resolved was developed and tested with RCNIC clinical managers and charge nurses. We predicted that manager and charge nurse understanding of Pyxis overrides and resolution process would increase. Feedback indicated education was appropriate, well received, and should be rolled out to staff at team meetings.
Members of the Pyxis override team attended team meetings and had “Lunch and Learns” to provide education to the staff. We predicted that education, coupled with increased awareness and resources, would increase the resolution rate to 80% within two months. The education was well received. It supports evidenced-based practice, aligns with the hospitals safety initiatives and was included in annual competencies. Resolution median increased from 33% to 70% after education was complete.
During the FMEA process, it was also identified that there was a discrepancy between Peripherally Inserted Central Catheter (PICC) order sets in EPIC and current clinical practice.
A change in clinical practice occurred on June 15, 2010, which included 1.9 to 2.6 French PICC lines in the RCNIC. The new lumen size now required flushes every six hours rather than the previous every eight hours. The current order sets built in EPIC were not revised to reflect the practice change. Also, there were multiple order sets for PICC lines and flushes.
The current process for ordering heparin flushes for PICC lines in the RCNIC was examined. A tool was developed to track orders and assure patients received the correct frequency of heparin flushes. An APN reviewed her caseload on two separate days to ensure that patients had the correct flush orders in EPIC. The hypothesis was that orders were incorrectly placed or absent from the chart. Auditing orders for PICC flushes was difficult. Staff did not have a clear understanding of appropriate flush orders. The tool was revised and implemented in the next stage.
The APN attempted to monitor flush orders for the entire APN team patient load. The prediction was that the audit would be effective in identifying current inpatients with missing or inappropriate flush orders. The appropriate orders for heparin flush with a 1.9 to 2.6 French PICC line became unclear. The orders could not be corrected at the time due to a lack of understanding of correct PICC flush orders. This audit was put on hold while data was collected from RCNIC nurses, the hospital-wide CVC (central venous catheter) team, and the clinical manager in charge of PICC line education regarding appropriate flush requirements.
As PDSA #5 resumed, heparin flush orders were audited during morning rounds by all APN staff. APNs and residents working in the RCNIC were educated on the new PICC flush guidelines. A copy of the guidelines was placed in the APN and resident tools. The guideline was utilized during daily rounds to ensure proper flush orders were in EPIC. As a result, heparin flush overrides decreased. The team worked closely with Information Systems to ensure that PICC flush order sets for the RCNIC population were updated in EPIC to reflect current practice. Information Systems and pharmacy worked collaboratively to update the house-wide PICC flush order sets to reflect the new policy. Once the order sets for PICC flushes were updated and reflected current policy and practice, a flow diagram was developed to support prescribers. After implementation of the standardized Heparin Flush Order Sets, a decrease in overrides for heparin was observed.
As shown in Figure 3, excellent progress was made. However, the goal of a 90% resolution rate was not achieved. This failure was partly due to focusing on interventions of lower reliability, meaning that these interventions are primarily dependent on human factors rather than system changes. For example, many times overrides were not resolved after emergency and controlled intubation medications were removed from Pyxis. There continued to be a gap related to nurses forgetting to link the override to the order before documenting the medication administration. A second barrier continued to be the override created when removing sterile water used to reconstitute medication. The issue occurred because there was not a process for the prescriber to order the sterile water. A mean of 70% resolution of overrides was achieved after targeted education to staff and prescribers in combination with heparin flushes reflecting current policy and practice.
Education provided to staff continued to be presented to all newly-hired nurses during orientation to the RCNIC. A job aid was available and posted at each bedside workstation. Continuing education was made available to the bedside nursing and respiratory therapy staff.
Override resolution rates continued to be tracked weekly by the Quality Outcome Manager and the data was presented monthly at staff meetings. The data was delineated to the specific medication(s) removed on override. It was noted medications that removed continued to be removed were emergency medications or heparin flushes associated with capping lines after stopping continuous infusions.
Residents received education during their monthly rotations, but there continued to be a gap in utilization of the PICC flush guidelines and practices.
Based on the outcomes reached in the RCNIC, the Project Administrator was able to implement successful changes on units with optimal outcomes. Using the principles of improvement as outlined above, institutions can adopt this process for a quality improvement project. Rapid improvements are possible by using these principles.
Early, C., Riha, C., Martin, J., Lowden, K., and Harvey, E. (2011). Scanning for safety: An integrated approach to improved bar-code medication administration. Computers, Informatics, Nursing, 29(3), 157–164.
Kowiatek, J., Weber, R., Skledar, S., Frank, S., and DeVita, M. (2006). Assessing and monitoring override medications in automated dispensing devices. Journal on Quality and Patient Safety, 32(6), 309–317.
Langley, G., Moen, R., Nolan, K., Nolan, T., Norman, C., and Provost, L. (2009). The Improvement Guide: A Practical Approach to Enhancing Organizational Performance (2nd ed.). San Francisco, CA: Jossey-Bass.
Paparella, S. (2006). Automated medication dispensing systems: Not error free. Journal of Emergency Nursing, 32(1), 71–74.
Rich, D. (2004). New JCAHO medication management standards for 2004. American Journal of Health-System Pharmacist, 61, 1349–1358.
Pamela Pockras is an advanced practice registered nurse at the Cincinnati Children’s Hospital Medical Center in the Regional Neonatal Intensive Care Unit. In addition to her NICU responsibilities, she serves on the Ohio Perinatal Quality Collaborative committed to decreasing blood stream infections in very low birth weight infants. She is also a certified Lactation Counselor and actively participates in multiple projects at both the local and state level to encourage the use of human milk and to support breastfeeding in at risk infants.
Reanna Smith is an ECMO Clinician at Cincinnati Children’s’ Hospital Medical Center with the Extracorporeal Life Support program. She has 12 years’ experience as a staff nurse in the NICU. She is NIDCAP trained and a certified Child Safety Seat Technician. She maintains an RNC certification as a high risk neonatal nurse and participates in many evidence based practice projects throughout the hospital.