Collaboration--integrating nursing, pharmacy and information technology into a barcode medication administration system implementation.

Abstract

Medication errors have been a national health care item since the Institute of Medicine's (IOM) sentinel report, which states that annually 98,000 people die in hospitals due to medical errors. The Agency for Healthcare Research and Quality (AHRQ) estimates that approximately 7,000 people die from medication errors annually and that an error occurs in 19% of all doses administered, 7% of these errors have the potential to result in an adverse effect.

Vassar Brothers Medical Center, a 365 bed hospital, implemented a Barcode Medication Administration (BCMA) system as the centerpiece of our overall patient safety initiative to reduce medication errors when ordering/transcribing medication; to standardize medication administration practices; and to bring administration documentation to the point of care.

Collaboration with Information Technology (IT), Nursing and Pharmacy for design, development, and implementation are critical. This collaborative process was viewed as a "3 legged stool" approach with each dependent on each other for stability and support. While collaboration among all levels must guide the process, without nurse participation at every juncture, any implementation stands a decreased chance of success. What determined our success was getting buy-in from those who would be using and working with this new technology. To embrace ownership, they needed to participate actively in the process. Our experience has generated considerations for developing and implementing similar applications. With the paradigm shift toward integrating more technology to care for patients, collaboration, good communication and teamwork are the heart of any organizational improvement.

Introduction

The Institute of Medicine's 1999 report, "To Err is Human," raised the healthcare industry's collective awareness of the critical issue of medication errors in the hospital setting. The direct effect of medication errors on patient safety and quality of care is enormous. Leape (1995) and his colleagues estimated that errors occur in 19% of all doses administered and that 7% of these errors have the potential to result in an adverse effect. These researchers learned that medication errors originate in the different phases of the medication use process in the following proportions: 39% during physician ordering, 12% during order transcription, 11% during medication processing, and 38% when a nurse administers a medication. Of interest however, is the interception rate. Over one-half of physician prescription errors and almost one-third of transcription and medication processing errors were intercepted, largely by nurses, before reaching the patient. Regrettably, only two percent of the nurse administration errors were intercepted before reaching the patient.

At a hospital like Vassar Brothers Medical Center (VBMC) located in Poughkeepsie, New York, where two million medication doses are dispensed annually, that translates to 26,600 potential significant errors. A study conducted in 2002 found that in-patients are subjected to about two medication errors every day (Barker, 2002 pg 1897). If we assume that only 1% of these errors result in an adverse drug event (ADE) and using the IOM's estimated cost of $4,600 per adverse event, the cost of such errors would result in a potential cost of $1.22 million annually.

Vassar Brothers is a 365-bed anchor hospital and the regional referral center for the three-hospital system known as Health Quest. The system is within easy commuting distance for approximately one million residents.

Five years ago, in an effort to assess the incidence of medication errors at the Medical Center and identify any trends in those errors, VBMC implemented a non-punitive policy of self-reporting medication errors. In 2005, nearly 250 errors were voluntarily reported. Without the benefit of technology to automatically capture medication error data, we suspected that this number reflected only a small percentage of the actual errors made annually.

Methods

To successfully maintain a high level of quality patient care, the hospital expanded our strategic vision. This included minimizing or eliminating patient identification and medication errors, reducing costs and improving patient safety. In early 2005, we began work on our Information Technology (IT) solution to improve our entire medication use process by implementing a medication barcode administration system. Barcode technology automates the "five rights" check by prompting the nurse to scan their name badge, patient identification wristband to access the patient's medication profile, and medication to verify that the drug, patient, dose, time and route all match. These checks are done at the point-of-care, at the patients' bedside, just prior to the actual administration.

The Executive team and staff at VBMC determined that nursing and pharmacy would be active partners in the implementation process. The driving force behind this decision was nursing 'ownership' of the medication record, and while pharmacy is responsible for the order entry information, it is primarily a nursing tool used for documenting medication administration. Therefore, nursing acceptance of the new system was critical to its success.

[FIGURE 1 OMITTED]

For our shared communication pathway we created a 'collaborative design team' comprised of staff representatives, the director of pharmacy, nurse managers, clinical coordinators, nurse educators, our informatics nurse, admissions personnel, and representatives from IT. Collaboration is "a process through which parties who see different aspects of a problem can constructively explore their differences and search for solutions that go beyond their own limited visions of what is possible" (Gray, 1989, p.5). Figure 1 is our concept of the "three legged stool." It was how we viewed the strength of support needed for the project. Pharmacy, IT and Nursing were project champions, equal partners who were dependent on each other to maintain balance. We had to exhibit Collaboration--Integrating Nursing, Pharmacy and Information Technology into a Barcode Medication Administration System Implementation mutual respect and have an understanding of the processes that would impact everyone's workflow. This was an important shift in the practice model. If the support is not equal, the stool can't support the weight of the project, others would need to pick up the slack. We believed that although it was a clinical system, to run efficiently and effectively using computers on our wireless infrastructure, we needed IT to be our 'third leg'.

[FIGURE 2 OMITTED]

Change of any sort is always challenging and sometimes very difficult. This planned change was a process that resulted from a well-thought-out and conscious effort to improve patient safety. It was more than just changing practice; we were adding work time to already stressed caregivers, NOT reducing it. Lewin's (1951) organizational change model is an appropriate theory to use in understanding the phenomenon of this dramatic cultural shift. According to Lewin, the driving forces are those that push toward a change position and the restraining forces are those that maintain the status quo. Therefore, in order to promote change, our driving forces for patient safety needs to increase while our restraining forces, staff resistance to change needs to decrease. To restructure our institution, we had to assist in moving through the three phases of organizational change. In Figure 2, we see how Kurt Lewin refers to the three phases as unfreezing, moving and refreezing. We, however, chose to call them project origination, workflow re-design and implementation and integration.

In the first phase, unfreezing or origination, we needed to concentrate our efforts on vendor selection, site visits, demonstrations and creating a Nursing Informatics position. We believed that one of the most common ways to overcome resistance and build support for change was through education. We included our users in making all of the critical decisions along the way. Device deployment is an important consideration when selecting and implementing a barcode system. The bedside devices are used to scan the barcode on the caregiver, the patient identification band, and the drug. We piloted five different carts and had over 500 evaluations rating each for mobility, stability, usefulness and ease of use. Our staff selected a portable bedside laptop computer on wheels (COW), because it allows the scanning process to occur near the patient. We had our carts customized with drawers for all the extra "stuff" that is needed to administer medications such as cups, pill crushers, alcohol pads, gauze, and band-aids. Our portable COW goes into a patient's room and the nurse can pull it up to the bedside so the patient can see the screen. The computer contains a wireless network card that facilitates real-time connectivity to the hospital's wireless system. The nurses can refer to the screen to answer any medication-related questions the patient might have, and the medication information button allows the nurse to link directly to the hospital's online drug information database. The nurse can access information about new or unfamiliar drugs and print out this information for the patient. "Equipment must be designed correctly or they can actually introduce errors. If nurses are satisfied with the system, procedures will be followed without "work-arounds," and the goal of decreased errors should be achieved (Grissinger & Globus, 2004 p.38). In the second phase, moving or re-design, we wanted to facilitate buy-in to the system, so we involved the staff in designing the product and its user interface screens for more than six months prior to implementation. A considerable amount of the design team's effort was spent tailoring the system to the unique needs of each unit. We met weekly to discuss, plan and design each module or part of the system. It was important that everyone involved understood how the decisions would customize the medication administration process and how it would change the present pathways. We had to change policies and incorporate best practice modules into nurse training before implementing our barcode system, in hopes of increasing nurses' familiarity with the technology and alleviating any concerns about performance expectations with the new system.

The biggest concerns among the staff were that the system slowed them down, took them away from patient care, and that standard hospital dosing times and "real time" administration would cause conflict. To alleviate some of these fears, we employed process mapping to show variations in how we communicate, how we validate and confirm orders, and how medications are delivered. By doing this, we often found a more efficient mode of practice, but there were a few instances in which medication administration was not performed according to our current hospital policy. We realized that although nurses employed these 'shortcuts' to save time, they were often unaware that they were jeopardizing patient safety. Looking at the workflows, the project team helped to identify opportunities to redesign the process and use the new technology more to maximize efficiency and improve patient safety. We could then align our policies and procedures with our new practice.

Pharmacy installed a robotic system for packaging medications with barcodes. This was a large project with joint effort by Pharmacy and IT. Although the robot would create any barcode format, it needed to be one that could be read by the scanners. Requirements for a hospital pharmacy-based drug-repackaging center include a high degree of automation; the capacity to handle a large volume; staff that are able to accurately perform multiple tasks simultaneously; back-up systems in case of mechanical failure; and a fail-safe verification process to minimize error (Neuenschwander, 2003). Pharmacists needed to understand how this would impact their workflow, change the way they interact with the floor nurses and most importantly, all of the transcription errors that were 'erased' with the ease of a pen in the paper world would not fit with the new system. They now would be visibly accountable, just like the nurses, for transcription, verification, dispensing and delivery of medications.

Long before our go-live, IT considerations revolved around the implementation of wireless local area network (WLAN) technology. They needed to consider the coverage areas, supported applications, point-of-care devices, infrastructure, and interference with other devices in the hospital. They played an important part in the hardware choices, setting up the devices, maximizing automated sign-ins and troubleshooting. Nurses do not tire of caring for patients; rather, they tire of dealing with inefficient systems. Technology enhancements must be able to help nurses focus their time on direct patient care. Our users wanted an easy-to-use, lightweight, and portable device to provide point-of-care data entry and a system that had the capability of accessing other hospital applications.

Lewin (1951) identifies refreezing (our implementation/integration) as the last stage of the change process. Our timeline for implementation took into account the time it would take to install all software and hardware, to train all users, and to make adjustments to the system as problems arose during initial start up. Staff training on the barcode system occurred over a three-week period immediately prior to go live. The training was conducted by our nurse informatics specialist, staff educators and trained super users. These super users then volunteered to assist in the unit training. After 6 months of intensive planning and preparation by a multidisciplinary team of more than 20 individuals, we were up and running in November 2005.

Daily meetings were held with the vendor, nursing staff, Pharmacy, Respiratory Therapy and our IT representatives during the first two weeks following implementation to trouble shoot any problems and to provide rapid response to questions about the system. In addition, issue logs were placed on each COW for nurses to write comments, provide information about unscannable items, and highlight any problems with the wireless network. Our informatics nurse reviewed these logs daily in order to address issues quickly. Management meetings were conducted daily with the Director of Pharmacy, several representatives from IT, and the Nurse Manager to ensure that the system was working properly. It took approximately seven months to implement our barcode system in all of the 15 nursing units, and two outpatient areas. We averaged about two weeks per nursing unit, with the first week for training and the second week go-live. Staff attended a 4-hour classroom session with hands-on learning on the computer, and each unit designated super users who were accessible on every shift around the clock for assistance and training. Unit Secretaries and Respiratory Therapists were trained with their own set of super users.

During implementation, we instituted a monthly user focus meeting. At these meetings we reviewed any problems and heard about fixes that individual units have utilized, pharmacy problems with 'bad barcodes', vendor changes, and meaningful feedback about identified issues. The meetings often led to a lively discussion, especially when we attempted to deal with compliance and workarounds. If a medication cannot be scanned, it is likely the nurse will employ a 'workaround' technique and select the medication manually, thereby bypassing the barcode system and placing the patient at risk for a potential medication error. To prevent these workarounds, we have identified reasons why they are occurring and areas where we need to retrain staff. We also have an edition of "Barcode News" that comes out monthly and is posted on every unit's Barcode Bulletin Board. This newsletter informs staff of the compliance rate, new changes and upgrades, troubleshooting tips and has a brief synopsis of the last user meeting events for those unable to attend.

Challenges and Improvements

Lewin believed refreezing was the last stage of the change process. But if by implementing new ideas and strategies they become our best practices, it is not the content of the change which we desire to "refreeze" but the learning processes--the attitudes, beliefs, skills and capacities which promote continued growth (Senge et al, 1994). Figure 3 demonstrates how our implementation has come full circle. In the beginning of this process, we believed integration to be just about Pharmacy and Nursing with strong IT support. But as we moved and changed, our vision and beliefs changed as well. With this hard won integration, we now experience a cohesive environment that shares the understanding of roles, expectations and limitations; agrees upon common goals and action plans; and strives for better outcomes for patient care. It is critical to note that this technology goes well beyond the prevention of medication errors at the bedside. Our experience has generated several points of interest for those undertaking such a collaborative endeavor.

[FIGURE 3 OMITTED]

Ownership

We realized during the course of our implementation that if we wanted to function effectively, ownership of the system must be a shared effort and responsibility. Pharmacy and Nursing created system administrators--but once we incorporated IT as a part of the clinical team, we found that they needed their own administrator. We often assumed that each part of our 'legged stool' understood their responsibility. Unfortunately, our IT department is not in-house. Vassar, as part of HealthQuest, outsources all of our technical expertise even down to our technicians. We continually needed to clarify the scope, milestones, ownership and responsibility for hardware delivery and setup.

Responsibility for troubleshooting, vendor contact, hardware problems and downtime support was shuttled back and forth, but mostly left for nursing to solve. Our Help Desk personnel were to be the first contact if a problem arose, but often they were ill equipped to handle more than minor problems. Together, we created a user guide for the Help Desk staff, including questions to ask the user and an escalation process, and we finally hired our own IT clinical technologist to coordinate all of the hardware tracking, problem solving and to be our vendor's IT point of contact.

System Optimization

Reliability is a significant factor in selecting clinical applications, because nurses depend on the system in critical situations. Although we had a newly installed wireless interface it took much 'tweaking' to ensure adequate coverage, especially in some of the far reaches of our older buildings. IT also had to ensure that this new technology was compatible for all applications, and they had to consider how the infrastructure would eventually need to be shared for additional or related applications. In addition to patient safety, reliable system availability and good connectivity can make a nurse adhere to policies and processes, thereby reducing unsafe 'workaround' practices.

Patterson et al (2002) identified negative consequences of implementation, including avoided activities (e.g., scanning wristbands to identify patients) to deal with heavy workload during busy periods. The researchers concluded that although the technology can help prevent some errors, the consequences may lead to new medication administration errors (Patterson, 2002). We addressed this serious workaround of not scanning the patient wristband as a severe breach of practice. Written warnings are issued to any caregiver found violating patient safety. To increase efficiency, use, and adherence to policy, we changed our label paper, replaced our scanners house-wide to those with a better-read area, and required wristband label replacements to be demand-printed rather than automatically generated.

Ongoing support by pharmacy is critical to making the barcode system work properly. Barcodes are tied to the national drug code (NDC) number for the specific drug that is maintained in the pharmacy system formulary. Pharmacy must ensure that new suppliers are added to the system properly so that the NDC is recognized as a valid barcode. Order entry by the Pharmacists needed to be more precise than was required before. The same drug that comes in tablet and capsule forms will have different numbers and different barcodes. Therefore if the order is written for tablets and capsules are dispensed, the barcode will not match the order. This mismatch was very problematic in the beginning for the nursing staff because they were used to administering medications based on patient needs, but now medications could no longer be interchanged because the barcode would not match. The biggest problem in Pharmacy was more work than expected. The system increased order entry time requirements for pharmacy. They were often asked to modify the scheduled administration times and nurses wanted to incorporate many non-drug reminders. One of the most beneficial features of the application in use at our hospital however, is "Pharmacy Messaging". This allows for real-time electronic communication between the nurse and the Pharmacist. They now have the ability to question an order and even ask for a replacement medication all without leaving the patient bedside.

Education and training were based on whether staff nurses had previous experience with computers. Nurses who were familiar with a keyboard and mouse learned the computerized barcode system more quickly. Those who had limited or no experience with either typing or computers required longer classroom instruction and additional follow-up support. We also identified two definite levels of learning. The first level is the ability to repeat a process successfully in a computer lab, although this does not always mean that the nurse will be able to follow the actions on his or her own unit. The second level is the ability to integrate the new process. In most inpatient areas, nurses access our barcode system via wireless laptop computers. Repeat training was identified as a method to ensure that nurses retained all information provided during their initial training. We also gave them concrete help in troubleshooting when problems arose with the computer. "Brush-up" classes were held in our training lab and on any unit where nurses continued to have problems.

Outcomes

We found that collaboration was critical to our success. Our barcode implementation has improved not only patient safety but also how our outsourced IT utilizes its resources and time. It has changed how they and the rest of the institution view their responsibility with hospital initiatives and they now understand that IT planning is not a separate process. In healthcare, when we rely on computers, they become part of the clinical environment just as indispensable as a stethoscope for patient care. Technology, as a result, cannot stay in the background when it comes to equipment choices, uses, and, importantly implementation.

At the outset, once the nursing staff was adequately trained and actually using the technology, the implementation team set a goal of 85% as our benchmark for unit compliance of medications the nurse actually scanned. Now that all systems are in place, we average over 90% hospital-wide and have set a goal of 95% as our new target. Why not 100%? That is not an achievable goal because there will always be non-formulary products and patients' own medications that are dispensed as part of the medication plan during hospitalization that simply do not get barcoded.

Barcode implementation caused a dramatic increase in the reported prevented or 'near-miss' medication error rate. In 2005 we had 255 self-reported errors. Figure 4 shows the number of prevented errors per month. Category C and D relate to the severity of harm that would have affected our patients. The National Coordinating Council for Medication Error Reporting and Prevention (NCC MERP, 1996) began using a medication error index that classifies an error according to the severity of the outcome. This index looks at factors such as whether the error reached the patient and, if the patient was harmed, to what degree. Our tracking of the Category C and D errors relates specifically to these NCC MERP guidelines. Category C is an error that occurred and reached the patient but did not cause patient harm, and Category D is an error that occurred which reached the patient and required monitoring to confirm that it resulted in no harm to the patient and/or required intervention to preclude harm. We prevent, on average, over 700 medication errors per month. The majority, over 60%, are those medications attempted to be administered far outside of our scheduled time parameters. Because of the alerts and warnings the nurses receive, we stop several hundred errors that have the ability to cause real harm to our patients.

Technology can provide a valuable tool for improving the safety of medication administration. Barcodes used to identify the patient and medication to be administered can provide an electronic double check of the 'five rights'. Through its effectiveness in displaying the patient safety benefits, data collection and feedback are one of the best tools to create early wins. We collect and display data that demonstrate improvements in patient safety. By providing valuable information, including some that might not otherwise be identified; this technology goes well beyond the prevention of medication errors. Just as importantly, the information it gives us can help to improve the medication use process. Our large number of prevented insulin errors (about one per day), for example, led us to re-educate nurses, pharmacists and our dietary department on insulin uses, administration timing and it also helped pharmacy identify and streamline our order process.

This technology helps us to go beyond our non-punitive self-reporting system for medication errors, and to begin to identify the total scope of potential errors and near misses. We have incorporated patient safety as part of our basic mission to improve quality, prevent medication and other errors, promote practice and have made it a commitment to our patients and the community.

[FIGURE 4 OMITTED]

References

Barker, K.N; Flynn, E.A et al. (2002). Medication errors observed in 36 health care facilities. Archives of Internal Medicine, 162. 1,897-1,903

Gray, B. (1989). Collaborating. San Francisco: Jossey-Bass.

Grissinger, M., Globus, N.J. (2004). How technology affects your risk of medication errors. Nursing, 34, 3642. Institute of Medicine. (1999). To err is human: Building a safer health system. Washington, D.C.: National Academy Press.

Leape, L.L; Bates, D.W; Cullen, D.J et al. (1995). Systems analysis of adverse drug events, Journal of the American Medical Association. 274, 35-43.

Lewin, K. (1951). Field theory in social science. New York: Harper & Brothers.

National Coordinating Council for Medication Error Reporting and Prevention. (1996). Types of medication errors. Retrieved May 8, 2007, from http://www.nccmerp.org/medErrorCatIndex.html

Neuenschwander M, Cohen MR, Vaida AJ, et al. (2003). Practical guide to bar coding for patient medication safety. American Journal of Health-System Pharmacy. 60, 768-79.

Patterson ES, Cook RI, & Render ML. (2002). Improving patient safety by identifying side effects from introducing bar coding in medication administration. Journal of American Medical Informatics Association; 9,540-553.

Senge, P., Kleiner, A., Roberts, C., Ross, R., & Smith, B. (1994). The Fifth Discipline Fieldbook. New York: Doubleday.

Related Websites:

* National Coordinating Council for Medication Error Report. www.nccmerp.org

* Institute for Safe Medication Practices www.ismp.org

About the Author:

Joanne Ross is a Nursing Informatics Specialist at Vassar Brothers Medical Center in Poughkeepsie, New York. She serves as the Nursing Project Lead on the Barcode Medication Administration Project. Joanne received her Master's degree in Nursing from Pace University with a concentration as a Maternal Child Clinical Specialist. She is also a graduate of the Post-Master's Certificate program in Nursing Informatics from Pace University, New York. She has also been involved in Pediatric research and National Institutes of Health (NIH) drug trials. Joanne is an active CARING member since 2005.

Joanne Ross, MSN, RN-BC

[e-mail]