Frush, Karen MD; Pleasants, Jane MS; Shulby, Gail RN, MA; Hendrix, Barbara; Berson, Brooke RN, MBA; Gordon, Cynthia RN; Cuffe, Michael S. MD
Many health care providers have implemented patient safety programs to reduce the potential of patient harm. Risk reduction strategies include prospective methods, such as Safety WalkRounds1 and Failure Mode and Effects Analysis; real-time surveillance using advanced information technology systems2; and retrospective risk identification and mitigation, often through a formal process such as a root cause analysis. The National Quality Forum recommends in “30 Safe Practices for Better Health Care”3 that these risk identification and mitigation activities be subject to enterprise-wide analysis to optimize pattern recognition and opportunities for systems improvement. On the basis of our experience at Duke Medicine, an organization that comprises an academic medical center, two community hospitals, outlying clinics, physicians’ offices, and home health and hospice, we believe that both patient safety and operational leaders must be cognizant of and manage risks that originate outside the walls of the health care organization: that is, risks related to the thousands of individual products that daily enter the health care supply chain.
Challenges of Managing the Recall Process
Hundreds of thousands of medical products are delivered to health care providers each week, including the continual introduction of improved or new medical products. At the same time, the number of product advisories and recalls, from both manufacturers and regulatory agencies, has increased by approximately 50% over the past several years. Product alerts are issued by the U.S. Food and Drug Administration (FDA) and manufacturers when clinical experience or ongoing research indicates possible or known safety exposures encountered in the use of the product. Such alerts may notify the provider of an update in software used in medical equipment, describe a new process for use of the product, or mandate the recall or changed use of a medical product or drug. Alerts are most often issued by manufacturers and may also be posted on the FDA Web site. Duke Medicine, through a Web-based subscription service (WBSS), received close to 3,000 alerts between July 2008 and June 2009.
The current recall process is fraught with challenges. Alerts are sent to hospitals through multiple and inconsistent channels, including mail, fax, personal communication from the vendor's representative to the customer, and e-mail; worse, these are directed at a host of potential end users and administrators. There are unique challenges tracing medical device use versus pharmaceuticals. Pharmaceuticals are encoded with a unique number in a standard format which makes them easier to track electronically. Medical device numbering, however, is not standardized, making medical devices difficult to identify in the supply chain.
Once an alert is received, the health care provider's goal is to route the alert to the appropriate individual responsible for taking action to resolve the alert. Depending on the product in question, this individual could be located in the catheterization lab, a department such as radiology or material management, or the pharmacy—nearly anywhere in the health care facility. Ideally, the alert is reviewed by the appropriate “responder” and a reply is completed and returned to the manufacturer, indicating the quantities of product involved for that health care facility and the actions taken to resolve the alert. Despite the best efforts of health care providers, weaknesses and gaps in the recall process lead to delays, fragmentation, and disruptions, thus extending the number of days patients may be at risk from potentially faulty or misused products.
Although patient safety is the primary reason for shortening the window of time for a recalled product to remain in the supply chain, the potential costs associated with medical product liability can be substantial. In addition to the impact on the patient, the hospital's reputation can be tarnished, resulting in loss of patient confidence and revenue.
Health care industry metrics indicate the average length of time for hospitals to acknowledge receipt of the alert, locate the product, and close the alert is 45 days (personal communication, L. Jeweler, Noblis). In addition to the inherent issues identified above, many health care organizations experience rogue buying patterns with products entering the health care organization at the point of use without organizational knowledge and approval. Multiple entry points for products create disconnects in product tracking, making effective recall procedures difficult.
To address these concerns, Duke Medicine implemented a WBBS in 2005 called the Risk and Safety Management Alert System (RASMAS). RASMAS is an advanced electronic “tool” that must be used appropriately by health care providers and safety leaders to be maximally effective. The purpose of this paper is to describe the implementation of a recall process using the RASMAS tool and to discuss the creation of a senior recall and response team that takes the computerized tool from being merely a source of information to an enabler for immediate critical decision making and communication.
Managing Product Alerts at Duke Medicine
In 2004, the drug and product alert system at Duke Medicine was a manual system housed in the pharmacy department. Alerts were gathered from various sources, compiled and sent out weekly, and tracked by number. The notification process involved a “broadcast” e-mail sent by the pharmacy department that was sent to approximately 500 employees, containing information about all recalls or alerts that had been received that week. Because the notifications were not targeted to specific product domains, recipients of the e-mails had to contend with a large volume interspersed with operational work e-mails. Each recipient then had to search his or her inventory independently to determine whether the product had been used and was on the shelf. If the affected product was found, an e-mail with the recall volume and follow-up actions was sent back to the pharmacy department, where it was printed and filed. When the product had never been purchased or was not stocked in areas covered by the alert, the expenditure of resources, time, and energy was frustrating for involved individuals and wasteful of resources for the entity. This paper-based filing and tracking process made information retrieval not only burdensome but inefficient, both in terms of documentation and monitoring.
Leaders at Duke Medicine recognized the need for a streamlined, centralized process that would facilitate the identification of recalled products as rapidly as possible after an alert was issued. Several senior leaders were introduced to RASMAS, a WBSS that provides comprehensive notification, distribution, and management of product alerts in health care organizations. This WBSS collects alerts nationally from numerous sources, such as the FDA, product manufacturers, and pharmaceutical companies, and sends them over the Internet to subscribers in a standardized format. Used to its fullest, the Web-based recall management tool provides subscribers with the ability to better manage alerts internally, whether related to medications, medical devices, or other medical supplies. After reviewing several other systems, Duke Medicine leaders elected to implement the WBSS tool throughout our health system.
Implementing an Improved System to Manage Product Alerts
To implement and fully optimize the new alert system, a tiered triage process was created to allow customized delivery of alerts, depending on the product involved and the department receiving the notification. The list of alert recipients was revised to include those individuals who order and receive products, because they are in the best position to quickly locate the products. Department directors assisted in identifying primary responders and establishing backup (secondary) responders. A first-line review and coordination function was centralized within the Procurement and Supply Chain and Clinical Engineering departments. As a result of that centralization and first-line review, searching Duke Medicine's enterprise-wide materials management system for a specific product became the responsibility of three recall coordinators, rather than the old manual system's broadcast message that went to 500 employees who were then burdened with ensuring their division or lab was not affected. The recall coordinators have clinical experience, thus providing an additional dimension to the triage of alerts which includes determining whether or not the item was purchased.
All primary and secondary responders who would be using RASMAS were provided with a brief online tutorial before the launch of the system in February 2005. The system was first rolled out to the three hospitals, with the existing manual process running in parallel across Duke Medicine's three hospitals, and then extended to outlying clinics and physician practices, including home care and hospice. Every entity within Duke Medicine is now included in the alert process. Although the automated alert system has more than 500 users at 81 separate locations, the users receive only alerts specific to their area of responsibility.
Impact of a Redesigned Recall Management Process
Within three months of implementing the WBSS tool, the time required to triage, resolve, document, and fully close alerts decreased from 43 days to an average of 2.74 days (see Figure 1). This improvement in response time has been sustained for three years, despite the fact that the alert volume has grown approximately 25% each year and has now reached 3,600 alerts per year. The increase in alerts is a combination of an improved capture rate achieved through the service and a rise in the number of alerts being issued.
Several key factors have contributed to the improved response time, including a built-in accountability feature of the WBSS tool and the implementation of enhanced processes for forwarding recalls received throughout Duke Medicine. All responders are required to close each alert. The system includes a notice escalation feature that signals managers and senior leaders if an alert is not closed. Lapses in response are rare, and an e-mail reminder generally results in resolution of the alert. Further, all responders receive guidance about the importance of arranging for backup responders when they are absent, to avoid a delay in response. In the event that anyone receives a notice outside of the WBSS tool, the individual forwards the information to the Clinical Risk Management Department. The risk manager who receives the information immediately forwards it to the recall coordinators, with confirmation of the completed delivery required. The recall coordinators include a pharmacist, registered nurse, and clinical engineer. The pharmacist and registered nurse are on staff in the Procurement and Supply Chain Department and have access to the enterprise-wide procurement and supply chain information system. The clinical engineer has access to the Duke Medicine equipment inventory. Keeping the process simple for users, while still ensuring accountability through the recall coordinators, is one of the keys to success.
The Senior Recall and Response Team
Another key success factor that led to the marked reduction in the average number of days to close critical alerts was the creation of a senior leadership team, which is responsible for the development of communication strategies and processes surrounding the Web-based recall management system.
Once the WBSS tool was chosen and then implemented across Duke Medicine, a Senior Recall and Response Team (SRRT) was established to evaluate the impact of product recalls and alerts on patient safety, to evaluate response action plans, and to provide oversight of patient and provider communication strategies. The team is small, comprising physician, administrative, patient safety, quality, risk management, and supply chain leadership, and is immediately convened when a recall, process change, or alert is believed to impact patient safety. The SRRT also meets on a scheduled monthly basis to track alert response action plans, ensure a closed-loop process has occurred, and address general issues and continual improvement process opportunities. Ad hoc invitations to leaders to attend one of these meetings are issued when the discussion and solution could benefit from their participation. For instance, the director of biomedical engineering was invited to attend a meeting at which equipment-tracking processes were being discussed. Leaders are also invited to meetings to report on the actions they took to resolve an outstanding issue.
When an alert occurs, the team evaluates information related to the alert and identifies both short- and long-term actions to be taken, along with needed strategies for patient communication related to a recalled product. For example, does the manufacturer recommend physician follow-up for patients exposed to a recalled product? Do physicians and/or patients need to be contacted? Is there a process change needed to ensure safety?
In the instance of a product recall directly affecting patients, it is vital that the communication with the patient be consistent and concise. A standard set of talking points is developed when it is determined that direct patient communication is necessary. These points are vetted by the SRRT before distribution to the appropriate members of the health care team. Specialists are consulted, as needed, for specific product evaluation, action plan design, and communication points. Direct patient communication is coordinated through the attending physician or, if appropriate, the Clinical Risk Management Department. In most cases, one informative letter can be sent to all patients. However, in some instances, patients must meet individually with their physician to assess the risks of the recall and to develop a plan for action.
Managing the Reimbursement Implications of Recalls
Billing and reimbursement implications of some recalls can be significant, especially when multiple follow-up appointments and tests are required. Recalls that have an impact on reimbursements, whether Medicare or private insurance, are now managed through an explicit policy. In collaboration with the SRRT, the Revenue Management Department of Duke Medicine devised an algorithm that includes a notification strategy and finance assumptions, to ensure a consistent process for managing all aspects of recalls. The Revenue Management Department works closely with the Clinical Risk Management Department to ensure appropriate tracking of impacted patients, disposition of charges, and resolution of accounts for each patient. We have found that a single, shared database for resolution of revenue management concerns creates the most effective communication base, providing real-time access to account status for both the Revenue Management Department and Clinical Risk Management Department. The process includes scheduled follow-up inquiries by the Revenue Management Department to ensure disposition on a timely basis while respecting the complexity that can occur with patients with multiple medical concerns. For example, it may be necessary to determine whether an implanted cardiac device is replaced because of a voluntary advisory notice or whether replacement was due to an unrelated patient issue.
An additional benefit to the working partnership is efficient forwarding by the Revenue Management Department's billing representatives to the Clinical Risk Management Department of concerns raised by any patient impacted by a recall. If a question is raised, the billing representative can take the information, provide the patient with contact information for the Clinical Risk Management Department, and post the concern to the shared database, which automatically notifies the Clinical Risk Management Department of a new matter for response (see Figure 2).
Lessons Learned From Duke Medicine's Recall System Overhaul
In designing a centralized system to manage risks related to medical product alerts and recalls, we have learned several key lessons that are demonstrated in the case examples (see Appendix 1). These include:
* The integration of a WBSS with a multidisciplinary team of health system senior leaders provided prompt and efficient evaluation and management of important safety issues and enhanced the integrity of the overall process.
* The need to shunt all product alerts (both internally and externally generated) to an automated system, so that the information can be effectively and appropriately disseminated and tracked.
* The importance of identifying a definitive group of individuals accountable for follow-up actions related to a recall impacting their area.
* Communication with physicians and patients affected by product recalls must be clear and consistent, and delivered promptly, in order to thoroughly and efficiently address any increased risk of harm to patients, while avoiding undue concerns and unnecessary diagnostic tests and treatments.
Although the fundamentals of recall management are complex, the timeline and actions taken must be swift and well defined. Communication lines must be clearly delineated and supported by a sense of urgency regarding product or pharmaceutical issues at all organizational levels. The SRRT provides the necessary cross-functional team approach to the evaluation, communication, and monitoring of vital product safety issues.
The operational costs to implement and manage a comprehensive recall process include an annual subscription fee for the WBSS tool and approximately 25% of the clinical recall coordinators’ time. The time devoted to the process by the senior leaders on the SRRT should also be taken into consideration. However, having the process centralized and streamlined, the SRRT can respond to the issues and concerns in a more effective and timely manner, thus eliminating much of the time previously spent attempting to coordinate a very complex process.
The medical supply chain poses an increasing risk to patient safety as the number of products used by health care providers rises and the volume of alerts and recalls continues to grow. We have learned the importance of designing a centralized system to manage and reduce these risks, including the critical need for effective communication among various members of the health care team and affected physicians and patients. Having a strong foundation for recognition, evaluation, communication, and process resolution enables quick responses to alerts and product issues. Whether the recall is internal or is received from an external source, clear communication points, along with support and direction from leadership to effect resolution, are critical to ensuring patient safety. Duke Medicine's unique infrastructure for alerts and recall management, along with oversight and leadership of the SRRT, has enhanced the effectiveness of the WBSS tool. Alerts are communicated effectively and responded to in a consistent and global manner. This comprehensive approach, which allowed the total processing time to full closure to be markedly decreased, is a critical component of a broader organizational strategy to improve the safety of our patients.
1 Frankel A, Graydon-Baker E, Neppl C, Simmonds T, Gustafson M, Gandhi TK. Patient Safety Leadership WalkRounds. Jt Comm J Qual Saf. 2003;29:16–26.
2 Kilbridge PM, Alexander L, Ahmad A. Design and implementation of a system for computerized adverse drug event surveillance and intervention at an academic medical center. J Clin Outcomes Manag. 2006;13:94–100.
3 Agency for Healthcare Research and Quality. 30 Safe Practices for Better Health Care: Fact Sheet. Available at: http://www.ahrq.gov/qual/30safe.htm
. Accessed August 7, 2009.
The Success of a Recall Process Enabled by Leadership Infrastructure and a Web-Based Management Tool: Case Examples
Case 1: Neupro (rotigotine transdermal system) recall
Issue. Specific batches of Neupro (rotigotine transdermal system)—used for idiopathic Parkinson's disease—were recalled because of possible impaired drug delivery leading to a worldwide shortage and an out-of stock situation in the United States. The recall affected 25 Duke Medicine patients who were prescribed the drug.
Summary of issue. In March 2008, a press release was issued announcing a recall of specific batches of Neupro (rotigotine transdermal system), because of the identification of rotigotine crystal formation in some patches. The existence of these crystals may result in the possible delivery of a dose lower than prescribed. Neupro is indicated for use in the treatment of signs and symptoms of early and late stages of idiopathic Parkinson's disease. This recall led to a worldwide shortage and, ultimately, an out-of-stock situation in the United States. In the press release, physicians were advised to wean patients from the drug to prepare for an out-of-stock situation.
Course of action. A sales representative notified the neuroscience clinic, a clinic within the larger private diagnostic clinic. The private diagnostic clinic patient safety officer (PSO) notified the Senior Recall and Response Team (SRRT), whose members then contacted appropriate personnel to confirm there were no recalled products in the inventories of the health system's three hospitals (for inpatient administration). The E-Prescribing outpatient database was queried, and patients and prescribers were identified. Additional patients were identified on the drug sample distribution logs by ambulatory clinic administrators. Lead physician prescribers were notified of the advisory, and suggestions for alternative therapies were obtained from consultants. Manual chart reviews were performed. The patient list, key talking points, advisory, and a sample patient letter were provided to the neuroscience clinic and prescribing physicians. Prescribing physicians notified patients in writing and by telephone.
Keys to the success of this recall response included comprehensive prescription and drug sample distribution record-keeping within the outpatient electronic prescribing database and clinic sample inventory logs. Electronic medical records and a strong communication system between PSO, SRRT, administrators, and providers were also important contributors.
Case 2: Pacemaker recall due to possible lead fracture
Issue. Because of possible lead fracture, pacemakers were recalled affecting over 800 Duke Medicine patients.
Summary of issue. Several notifications from the manufacturer in October 2007 evolved from a simple warning bulletin into a full-fledged recall, with the latest advising physicians to stop implanting these leads and return the unused inventory. The company had estimated that 2.3% of the specified leads could fracture within 30 months of being implanted, as compared with the 0.9% rate for other leads. When lead fracture occurs, it can cause the device to deliver a powerful, unnecessary shock to the patient. It may also cause the device's battery to prematurely drain. According to the manufacturer, there were associated deaths and complications that had occurred. Because leads are positioned in the heart and tend to have scar tissue build up around them, the risk of removing a lead can outweigh prophylactic replacement.
Course of action. On receiving the alert, the end users were instructed to remove these leads from inventory. Simultaneous notifications to the SRRT and the patient revenue management organization (PRMO) were issued. Within one day, the SRRT was provided with a list of Duke Medicine patients affected by the alert. The SRRT, senior medical leadership, and providers drafted patient correspondence. This approach allowed a consistent message to our patients regarding their recommended follow-up care. The patient letters were sent within the week, the affected items were deleted from the material formulary, and the alert was closed within 10 days of initial notification.
After notification, the PRMO managed the billing aspects of the process. Per protocol, the PRMO received the patient list, records were flagged, and a PRMO task force was convened. Cited Here...