Minimizing Medical
Product Errors

---

A Systems Approach

We would like to express our gratitude to the FDA planning group which has been considering the issues associated with Minimizing Medical Product Errors-A Systems Approach including: Mr. Peter Carstensen, Mr. Jay Crowley, Dr. Elaine Esber, Dr. Mary Fanning, Dr. Steven Goldman, Ms. Mary Gross, Ms. Sharon Smith Holston, Ms. Areta Kupchyk, Dr. John Matheson, Mr. Thomas McGinnis, Dr. Stuart Nightingale, Mr. Jerry Phillips, Ms. Cornelia Rooks, and Dr. Susie Sensabaugh. We would also like to acknowledge the speakers who participated in the workshop including: Ms. Sharon Smith Holston, Dr. Janet Woodcock, Dr. David Feigal, Dr. Bruce Burlington, Ms. Diane Kennedy, Ms. Diane Cousins, Dr. Richard Cook, Dr. Lucian Leape, Mr. Mike Cohen, Dr. Bruce Lambert, Dr. James Battles, Mr. George DiDomizio, Dr. Dan Raemer, Mr. Steve Smith, Dr. Alan Goldhammer, Dr. Joseph Cranston, Dr. Sam Kidder, Ms. Linda Golodner, Mr. Thomas Granatir, Dr. Paul Minicozzi, Mr. John Santell, Mr. Ken Kopesky and Mr. Daniel O'Neal. Special recognition is given to panel moderators: Ms. Sharon Smith Holston, Dr. Bernard Schwetz, and Dr. Bruce Burlington and to Dr. Stuart Nightingale, who facilitated the Open Public Hearing. Finally, FDA wishes to thank the following people who commented during the open public hearing: Dr. David Classen, Dr. Joshua Schwartz, Mr. Albert DeRichemond, Mr. William Ellis, Ms. Eleanor Vogt, Dr. Neil Lerner, Dr. Bruce McWhinney and Dr. Dean Gubler.

Ms. Sharon Smith Holston, who is the Deputy Commissioner for External Affairs, FDA opened the workshop by saying that some 1.3 million people are accidentally injured by medical therapy in the U.S. and that the misuse of medicines cause thousands of preventable hospitalizations every year and results in estimated costs ranging anywhere from $20 to $75 billion. But more important than the financial cost associated with errors in medicine, are that the consequences of a medical product mishap can be serious and even life-threatening to the patient. She said that FDA knows that a problem exists, but that we don't know the true extent of the problem, or whether the problem is getting better or getting worse. FDA does believe, however, that the number of reported incidents is vastly underestimated because most of the reporting is done on a voluntary basis. She said that the goal of the workshop is to consider how FDA should use its' regulatory influence to reduce the risk to all Americans from misused therapeutics. Ms. Holston also acted as moderator of Panel 1. Panel 1 speakers included experts from in and outside government who shared their experiences relating to medication, biological product and device errors. Dr. Janet Woodcock, who is the Director, Center for Drug Evaluation and Research, FDA spoke first.

FDA Center Perspectives on Error

Dr. Woodcock spoke about medication errors from the Center for Drug Evaluation and Research (CDER) perspective and said that errors are not the result of an individual failure, but of a systems failure that results from complex interactions within the health care system and that with proper safety designs of products, we can decrease the potential for injury from the products regulated by FDA. Medication errors can occur anywhere in the delivery of a drug to the patient, including during the prescribing, repackaging, dispensing, administering of the drug and even while monitoring of the patient. Some causes of medication error are poor communication, ambiguity in the product names, problems with directions for use and problems with abbreviations. Along the continuum, any health professional may contribute to real or potential errors and sometimes, changing the design, the name or packaging of the product can prevent medication errors. When a medication error occurs, the risk manager of the hospital examines the system within his/her own purview and takes preventive steps. Often, however, discussions of these incidents are not shared with other hospitals, health care professionals, FDA or the drug's manufacturer. She said, that by not sharing these experiences, we are losing an opportunity to prevent similar occurrences.

FDA has received 6000 medication error reports since 1992 and 50 percent are related to confusion in the labeling or packaging of the drug. The total number of serious adverse events that are related to errors number 1,273 with 326 deaths. These numbers only represent a snapshot of what is actually occurring and FDA believes that there are a correspondingly greater number of unreported events. Currently, FDA has documented that errors can be traced to 441 cases resulting in patient hospitalizations, 235 cases where the patient's life is threatened, 206 cases where patients undergo medical intervention and 65 cases where patients experience permanent disability.

What is CDER doing now to prevent medication errors? First, prior to a drug's approval, CDER's Labeling and Nomenclature Committee reviews the proposed proprietary names to minimize mixups with existing names. In addition, CDER monitors reports of medication errors through the Medication Errors Committee and takes appropriate action. CDER also provides feedback to health care professionals about medication errors through publications in FDA's Medical Bulletin and the FDA CONSUMER magazine and collaborates with USP and AMA's Institute for Safe Medication Practices. Dr. Woodcock said that in order to effectively minimize the incidence of medication errors, CDER should focus on preventing the errors relating to product package and labeling prior to approval and correcting the problems that occur after approval. She said the pharmaceutical industry should propose clear and concise labeling and avoid trade names that are misleading or confusing and that the health care team should provide clearly written prescriptions and that practitioners should report the errors so that we can share the knowledge with manufacturers and others.

Center for Biologics Evaluation and Research (CBER)

Center for Devices and Radiological Health (CDRH)

The next speaker was Dr. Bruce Burlington, Director for CDRH, who said that problems seem to be reported more frequently with pharmaceuticals than they do with devices, but that he wasn't sure if it could be attributed to current reporting mechanisms or whether it was representative of reality. Within the device arena, more problems are reported during vascular and cardiac surgery and neurosurgery than in other areas. He said that FDA has spent considerable time to ensure that products are well designed and that the agency and manufacturers have achieved a moderately high level of success. However, CDRH has analyzed the incidence of device errors and discovered a significant number of product selection errors, that is, where the product is not matched appropriately to the patient needs. Device errors may be attributed to human factors in the the use of the product, design features of the device in terms of those human factors, our ability to communicate properly with the user about how the product should be used; the interaction between the level of training in the devices and the extent to which devices are designed for intuitive operations. He said he was struck by the lack of knowledge in this area and that he was sure the incidence of errors were under reported.

He also said that we do not have an active surveillance system to facilitate learning when device errors are occurring and against what background and what denominator which is an important tool to help resolve these issues. FDA is keenly aware of the value of appropriate exposure information--or denominators in assessing the frequency of errors, but there is no new mechanism to obtain the denominator of patient exposure with devices. Dr. Burlington said that CDRH had decreased the number of device associated injuries by performing a more thorough review during the premarket approval phase of the device. Over 1997-98, FDA has by regulation, implemented a requirement for device manufacturers to perform a design control analysis and part of the analysis is evaluating the operating features of the device prior to approval of the device. This requirement, which will be implemented next year, has the potential to significantly reduce the potential for errors occurring in devices. Dr. Burlington also touched on current efforts to improve postmarketing surveillance to better monitor adverse events occurring with devices. He said, coupled with CDRH's need for denominator data, that we should develop a system where FDA can actively look for problems related to those exposures, in order to identify the patterns of misadventure that sometimes results in patient injury. CDRH has established a pilot program in some hospitals and hopes to ultimately gain a better information base from which to approach this issue. CDRH is also encouraging better use of systems analysis in order to recognize that there are many participants involved who could, during one step or another, contribute to the user error, including the patient, the health care provider or operator and the device itself. He said that the problem of errors occurring with medical devices is compounded because we are dealing with many different devices, the instructions on the devices are not clear nor are they readily available to the operator, the design features are not constant and there are cost considerations as well as societal considerations revolving around the way operating room practices occur. In summary, Dr. Burlington said that FDA is concerned about the use of devices that could potentially harm patients and that FDA's involvement shouldn't stop at the time a product is approved. When device errors occur, we need to use a systems approach to analyze when bad results occur, when to intervene and how best to monitor those interventions.

Voluntary and Mandatory Reporting

Dianne Kennedy, who is the Director of FDA's MedWatch program, spoke on FDA voluntary and mandatory reporting requirements. She said that FDA has regulatory responsibility for ensuring the safety of all marketed medical products. She cited safety concerns to be: adverse reactions and product quality or use problems, including medical product errors attributable to confusing labeling, packaging design or similarities among product names. Some reporting is mandated by law or regulation and other reporting, although vital, is strictly voluntary and all reports begin with the single doctor, pharmacist or Medical Product Reporting Program. (MedWatch), which is a promotional outreach initiative, is designed to educate health professionals about their role in postmarket surveillance of adverse events and to facilitate reporting to FDA either directly or through the product manufacturer. The MedWatch form is used to report both adverse events and product problems with almost all FDA regulated products. Confidentiality of the reporter, as well as the patient, are protected through the MedWatch program. It is especially important to report an adverse event to FDA that is fatal, life-threatening, permanently/significantly disabling, requires or prolongs hospitalization, a congenital anomaly or requires medical or surgical intervention to prevent permanent impairment or damage. Many of these reports have the potential to be associated with a medical product error.

Manufacturers, distributors and labelers of drugs and therapeutic biologics must report all adverse events of which they are aware. If the event is serious and not included in the product labeling, it must be reported within 15 days and all other adverse events would be reported periodically. Manufacturers must also submit an FDA field alert to the appropriate FDA field office within 3 days where drug labeling may be mistaken for another article. Reports of adverse events with vaccines are sent to FDA using the Vaccine Adverse Event Reporting System (VAERS), which is a joint FDA/CDC program. The manufacturer of blood or blood products must promptly notify FDA of all errors or accidents that may affect the safety, purity or potency of any product, and when there is a complication.

The Safe Medical Device Act of 1990 requires that user facilities, that is, hospitals and nursing homes report deaths and serious injuries to FDA. Deaths must be reported to FDA within 10 days. Manufacturers of electronic products must immediately report all accidental radiation occurrences that arise from the manufacturing, testing or use of the products and they must also notify FDA of any product defect or failure to comply with applicable Federal standards. All reports, voluntary and mandatory, are routed to the appropriate FDA program office and each program has procedures for identifying, evaluating and taking action on any reports that may have resulted from medical product error.

Diane Cousins, who is the Vice President for the Practitioner Reporting Program at the United States Pharmacopeia (USP) was the next speaker. Ms. Cousins began her talk by discussing USP's experience in practitioner experience reporting and medication errors. The USP involvement began in 1991, when it became clear that medication error information would be useful in the development and revision of USP-NF standards and USP-DI warnings. The focus of the program is not to identify who made the error, but rather to take a systems based approach by focusing on the causes of errors and analyzing the circumstances surrounding the event. USP has reported that nearly a third of the reports are cases of potential error, that is, near misses, warnings, or opportunities for error which could, under the right circumstances, lead to inappropriate drug therapy and possible patient harm. Of the errors actually reaching the patient, approximately one-fourth resulted in temporary patient harm or permanent impairment; 6 percent were fatal. Injectables were reported as the dosage form most often involved in serious and fatal errors. Over the last year, the most common types of errors reported were medication errors and cases where the wrong drug was prescribed, dispensed or administered. The top 10 causes of errors include: similar labeling and packaging (1), look alike brand names (2) and sound alike brand names (8). Labeling and packaging problems accounted for 35 percent of reports last year, usually due to confusingly similar trade dress among products or similar packaging of the different strengths of the same product. Because USP reports show similar labeling as a frequently reported cause of error, error prevention efforts to differentiate product labeling for the most frequently reported drugs causing fatalities should be employed by companies carrying such a product line. Ms. Cousins closed her remarks by saying that she believed that the time was right to recast product safety evaluation in the pre-approval stages to include a sort of simulation testing for packaging, labeling and nomenclature, and errors should no longer be an excuse to blame, but rather an opportunity to improve medical products and the systems in which they are used.

Functions A'la Mode: Sources of Operating Failures in Microprocessed Based Medical Devices

Dr. Richard Cook, a physician, educator and researcher on error at the University of Chicago was the next speaker. He opened his talk by saying that the study of infusion devices is a model of the study of technology applications in medicine. Failures associated with infusion devices result from their complexity, in combination, with the work environment and human operator cognition. Because these failures are complex and hard to characterize, it is often difficult to correctly extrapolate what lessons should be learned from these failures. When viewed cognitively, specific meaning and implications can be drawn from the failures including:

1. Failures are shown to be mismatches among human operator cognition, device features and characteristics of the user environment;
2. The complexity of the device design is a critical factor relating to device failure;
3. Device design complexity is not reduced by concealing it under superficially simple interfaces;
4. Attempts to eliminate "human error" by using automation are ill-conceived and generate problems for operators that result in new kinds of device failures;
5. The underlying problem cannot be overcome by increased training.

Dr. Cook described a set of infusion device studies which exposed several different flaws resulting from user interaction with the device. Some flaws were classical human engineering deficiencies, but the most important ones were subtle but became evident during the detailed field study of the device and its use. These findings culminated in a design project that used a more direct understanding of the operation of the device. This research implies that "errors" associated with devices neither indict the device users or the device manufacturers, but rather act as markers for mismatches between device and world complexity.

Studies also show that device incidents are complex events and not easily summarized or categorized, especially with regard to whether or not "human error" played a role in the incident. Categorizing accidents as device failure or user error misrepresents how circumstances leading to failure can occur. Accidents involve a combination of individual minor factors that join together to create the potential for failure. What appear to be minor device design characteristics when viewed individually, can clearly become multiple critical factors that could lead to user error in certain settings. Device operators recognize and adapt to the perceived risks posed by devices, circumstances and procedures and anticipate forms of failure in an effort to prepare to cope with failure. In this, they are constructing a safety net in response to the perceived hazards of the workplace. This process of constructing a safety net to what are perceived hazards calls into question whether the term "error" is a simple, easily understood and atomic event. Rather, Dr. Cook said, what we call "error" is the end result of the complex, dynamic process of recognizing and adjusting to hazards while trying to meet other goals. Dr. Cook used modes and "mode error" as a model for problems that occur with infusion devices. He said that the demand for more automated features in devices results in the manufacture of devices with more internal states or "modes". The device complexity required to handle these multiple modes and their associated faults and error conditions necessarily makes devices more complex. Failures occur when the device interacts with the user in one mode, but when the operator expects that it will be in another. Because operators work in a multiple task, event-driven world, the opportunities for such mode errors are frequent, especially when devices camoflauge their modes or change modes "automatically". To reduce the potential for overt failures occurring requires a user centered approach to design. The redesign project illustrated one means of modifying device design by reducing the number of modes and allowing the device operator to see "into" the device more effectively. While these improvements are, in principle, available to all device manufacturers, it will require substantial effort to achieve this goal. Most importantly, it requires a deep understanding of the effect on human operator cognition of the device and how the device operates in the field. Failures associated with devices are but one means for exploring the relationships among operator, device characteristics and environment in new and revealing ways and as such, represents a good starting point for reducing the incidence of user error with medical devices.

How Can Manufacturers Help Hospitals Reduce Medication and Device Errors?

Dr. Lucian Leape, an adjunct professor at the Harvard School of Public Health and a member of the Health Sciences Division at RAND was our last speaker in Panel 1. Dr. Leape said he thought that drugs were inherently safe and that there seemed to be very few serious drug side effects, thanks to FDA and the drug manufacturers. But even so, in practice, he said, we still experience significant problems because doctors and patients don't use drugs right, and doctors, nurses, pharmacists and patients make mistakes. He asked what we can do about these errors, why they occur and how serious the problems were? He defined adverse drug event as any injury related to the use of the drug, whether or not it was due to an error, and in practice, less than half of them are due to errors. But he said, what we know about adverse drug events depends on how we look at these events. In most hospital reporting systems, we've seen that there are approximately two incident reports related to adverse drug events filed for every thousand admissions to a hospital, making the rate of incident reports to be 0.2 percent in terms of admissions to hospitals. But active investigations found that 6.5 to 10 percent of patients experience an adverse event, and one-half of those adverse events were attributable to errors. He said that a fundamental problem is that we rely on perfect performance and a belief that the way to eliminate errors is to train health professionals and then expect they will not make mistakes. He said the most effective way to minimize errors is to redesign systems using these human factors principles:

1. ) Reduce reliance on memory,
2. ) Simplify to the extent possible,
3. ) Standardize (e.g., doses, preparations, times etc.),
4. ) Use constraints and forcing functions.

He said we have multiple actors, who participate in the medication system alone, thereby creating a very complex system and our approach to reducing error seems to center almost entirely on the concept of developing and enforcing a series of rules. He said that doctors and nurses, and to some extent pharmacists, don't have the necessary information at hand when they are prescribing and distributing medicines and often depend on memory for dose/time and medication interactions. In addition, stylistic practices continue to be accepted in the medical profession that wouldn't be tolerated elsewhere. He said we were dealing with rare events, however, in the aggregate, they were very significant because people are dying and because it makes it difficult for anyone to declare ownership of the problem. In addition, he said that there is still fear of punishment because it is still widely believed that errors are human failings which can explain why only 0.2 percent of incidents are reported instead of the possibly 10 percent that are actually occurring.

He asked the manufacturers to make the following high leverage changes prior to marketing the drug: provide medication in unit dosing, bar code medication, use incompatible connectors to ensure that injectable medications are not misused, increase automatic dispensing, and provide pre-mix IV solutions. He also encouraged the manufacturers to simplify the labels in terms of concentration. He said low leverage changes are those changes which don't affect a great number of people but can still be crucially important and gave as an example the misuse of concentrated potassium chloride which kills people, but only rarely. Similarly, labeling and name confusion problems may rarely cause problems, but in the aggregate, the numbers add up. They are serious, and consequently should be addressed. Dr. Leape closed by asking manufacturers to invest one percent of their research and development budget to study the problem of medication errors.

Questions and Answers - Panel 1

A question was raised about whether the current tort system had a dampening effect on accurate reporting on adverse events that could be linked to errors. Dr. Leape responded by saying that according to his studies, fewer than 2 percent of people who received care resulting in negligent injuries ever file a suit, which in retrospect makes it a very ineffective way of providing compensation. He recommended that there be enterprise liability for medical injury; and that the hospital be responsible for compensating people when they are injured. He said the hospital would then have some very strong incentives for improving safety and reducing injuries. Dr. Feigal added that while we are considering tort reform, we should consider what the economic incentives are and how the economic incentives will add costs to the current system.

Dr. Leape was also asked to speculate on possible incentives that could be directed towards pharmaceutical manufacturers to encourage them to adopt the practice of unit dose packaging. Dr. Leape said that he thought that the drug companies did not always depend on profit making as the bottom line, and that at times, they perform altruistic acts just for the sense of doing good. He said that if the costs can be passed on, it's not detrimental to the companies and it's good for the public health, why not do it?

There was a request for elaboration on FDA's regulatory role, including whether FDA should consider that by restricting the use of certain devices according to packaging, was in fact, regulating the practice of medicine? Dr. Feigal responded and said that FDA does not regulate the practice of medicine including off-label indications of medical products. He said that a product should be safe and effective and promoted for the indications labeled and that FDA only regulates the promotional activities of the sponsors. Before a product is approved, the FDA sets the standards for a medical product that will be widely used in the general population.

Dr. Leape asked why package inserts are not more clearly written so they are universally understood? Dr. Woodcock agreed that current package inserts sometimes do not communicate crucial information to practitioners effectively and that FDA has held some public meetings in order to develop inserts that consistently emphasize the most important product information.

Panel 2 was composed of six outside experts who submitted scientific abstracts to FDA on possible methods that already exist that can help assess the potential for medical product errors. Panel 2 was moderated by Dr. Bernard Schwetz, who is Director of FDA's National Center for Toxicological Research and interim Chief Scientist. The first speaker was Mr. Michael Cohen who is President of the Institute for Safe Medication Practices.

Failure Mode and Effects Analysis - A Way to Eliminate Medical Product Related Errors

Mr. Cohen described failure mode and effects analysis as a systematic process of understanding how and where errors might actually occur during the development phase of the product. He said this process borrows expertise from front line practitioners and guidance from experts in the field in order to succeed. He said that medication errors result from knowledge and performance deficits, that these deficits happen to everyone and can be traced to stress in the workplace and an inordinate amount of routine interruptions. He said that we need to develop a system to prevent these errors from translating into medical accidents where a patient is actually harmed. He said the role of industry is to properly design the drug package, label and name to help attenuate an error, prevent it entirely or to minimize the consequence of the error after it's happened. He described failure mode and effects analysis as having many variations and identified the following steps as being part of the process:

Mr. Cohen said that a systematic assessment of how and where systems might fail should be undertaken and where it is found that the effects of error are intolerable, action should be taken to minimize the potential for errors, make them visible or minimize their consequences. In failure mode and effects analysis, the focus remains on the system and process and not on who makes the error and on preventing errors from happening rather than reacting to them after the fact. Based on ISMP experience with practitioner reporting of medication errors, Mr. Cohen suggests the following actions to minimize errors:

1. FDA should establish minimum standards for systematic pre-launch qualitative testing of labeling, packaging, trademarks, and devices.
2. FDA should develop standards for independent outside organizations that utilize a system of pre-launch review by front-line practitioners with appropriate expertise in failure mode and effects analysis.
3. FDA should require companies, at the time of application, to certify that required error prevention testing to assure safe labeling, packaging, naming, device design, and instructions for product use, including appropriate warnings as necessary is complete.
4. Documentation should indicate how the testing was done, the outside agency that assisted, and should indicate that the problem was free from significant flaws or should indicate how problems that were revealed were resolved.
5. FDA should reserve the right to see the full documentation and to interact with the company and consulting firm when necessary.

Safety Risk Management in the FAA

Mr. Steve Smith spoke next from the Federal Aviation Administration. Mr. Smith drew similarities between maintaining aviation safety and maintaining patient safety and said that many of the techniques that are used to help understand and improve patient safety could be similarly applied to aviation. He said that generally, it's just not a single event that could lead to a major plane crash. He additionally stated that how you examine risk, identify failures and come to mitigation strategies is the same whether it applies to medicine or aviation safety and in order to minimize the potential for problems, it is necessary to plan, identify, analyze, assess and then decide how to either eliminate and/or mitigate the hazards. He defined a hazard as a condition, event or circumstance that can lead to an unplanned or undesired event. Risk was defined as nothing more than the product of the probability of that hazard occurring multiplied by the severity of that particular hazard. Certain factors remain finite that the FAA works around in order to ensure air safety. For example, over the next 15 years, more people will want to fly, putting more aircraft into the system, while the number of airports remain constant. It is impossible to extend the number of hours in a day to accommodate the extra number of travelers; therefore, the FAA needs to explore how to increase capacity within the existing air traffic system while maintaining or improving the current safety level. In contrast, if the FAA allowed the incident/accident rate to remain at the current level and allowed increased capacity at the anticipated rate cited above, within the next 20 years, we would be faced with the possibility of one major air carrier crash a week. This scenario is, of course, unacceptable and the FAA implemented a Strategic Plan with a goal of "Zero Accidents", a pivotal component which was safety risk management. The FAA currently holds cross-organizational meetings to discuss how a multi-disciplinary system would evolve and during these meetings, they assess airspace, workload and the state of controller equipment. Mr. Smith stated you can eliminate hazards but cannot eliminate risks without eliminating the hazard. It is possible to mitigate risks, that is, to decrease either the occurrences (the likelihood) or the severity of the event if it does occur. Another possibility is transferring the risk, which he described as not a viable option, because ultimately someone has to take responsibility and pay.

Mr. Smith stated the risk process is five steps that includes:

1. Plan and make a concerted effort to understand the issue under evaluation,
2. Identify your focus clearly,
3. Analyze by evaluating the worst case scenario and describing the severity of the event and the probability of occurrence of each of the hazards,
4. Assess the results and identify the "hot spots", and
5. Manage the risk, make the decision.

After undertaking all five steps, one should consider the need to eliminate the hazard, mitigate, transfer or accept the risk. Mr. Smith said that after you believe the problem is resolved, the process should be repeated to ensure that you've not missed anything of importance or have indeed acted upon the risk appropriately. During the question and answer period, Mark Bruly from ECRI, asked Mr. Smith to contrast and compare the number of different technologies in the medical device area to the different technologies within aviation. Mr. Smith said that the the two fields were similar. In fact, the FAA reviews some personal or portable electronic devices that travel in aircraft to see how they may or many not interfere or interject with some of the systems onboard the aircraft. Mr. Smith offered to discuss this further with Mr. Bruley at the conclusion of the panel.

A Medical Event Reporting System in Transfusion Medicine

The next speaker was Dr. James Battles, who is an Associate Professor in the Office of Medical Education, Southwest Medical Center. Dr. Battles said that rather than concealing our errors, we should explore them, in an effort to improve our systems. He also touched on the need to study benign errors in order to learn how to minimize errors before adverse events occur. Errors are classified into different groups and the active error is the error committed by those in direct contact with the human system interface (e.g., health professional). He cited an example of where one death occurs in 600,000 blood transfusions that is related to complications of A/B/O incompatibilities. He also said that we have 1 in 30,000 A/B/O incompatible blood transfusions that did not result in deaths and 1 in 12,000 units that were incorrectly hung. So, problems of a near miss nature are much more frequent and problematic than when a fatality results that can be traced to incompatibility in blood transfusions. He said the purpose of any event reporting system is the prevention of future errors and adverse outcomes. One key element of the near miss reporting system is there is no blame assigned to the reporter. It was determined that the most efficient way to get a transfusion system operating was to integrate it into already established quality assurance activities (like blood banks and hospital reporting systems). Dr. Battles used a technology called "smart paper" which combines optical character recognition and optical mark recognition which allows the person who completes the form to also be the data input person. The system incorporated the following information: identification of the event, where in the system the event occurred, which type of employee experienced the event, why did the event occur. This system then describes, classifies, computes and interprets the data. Detection describes where and when the event occurred and if you can't learn where in the process, it can tell you how many barriers were breached before this event was captured. Dr. Battles said that if you switch to a near miss reporting system, if there is a no fault mechanism and reporting is encouraged, there should be a dramatic increase in the number of reports to the system. It is then possible to profile the causes as well as the types of events that are occurring prior to the time actual harm comes to the patient.

Simulating Crisis as a Means of Evaluating a Medical Device

The next speaker was Dr. Dan Raemer who is Associate Professor of Anesthesia at Harvard Medical School, Bioengineer in the Department of Anesthesia and Critical Care at Massachusetts General Hospital. He discussed a program called crisis resource management which was designed to demonstrate to physicians how they might react when a medical crisis unfolds. Dr. Raemer began his presentation by showing a videotape that demonstrated an obstetric airway crisis in a pregnant patient. The crisis worsened because anesthesiologists did not recognize that the anesthesia machine switch was turned off. Dr. Raemer and colleague, Dr. Feinstein, observed other incidences where medical staff were not aware that the anesthesia machine was in the off position. Based on their videotape review, one contributor was the design of the anesthesia machine on/off switch. From this initial review, they followed up with a small study at three institutions, where they asked 164 anesthesiologists if they could recognize the switch from a pictograph. 90 out of 164 anesthesiologists correctly identified that the switch was in the off position, which statistically, is no different than a 50-50 guess. He said he thought that the technique of realistic simulation can be used in a variety of ways to study errors in medicine and particularly to pick up product errors in their context. He said it was a powerful tool for discovering errors that otherwise wouldn't be noticed; for evaluating and testing devices once they're produced. A realistic simulation is repeatable, efficient and a safe modality. No patient harm can come from this approach and it addresses the issue of a systems approach to medical error because the environmental issues that confront a clinician in real time are present.

Automated Screening for Look-Alike and Sound-Alike Drug Names

The next speaker was Dr. Bruce Lambert, who is Assistant Professor, Department of Pharmacy Administration, University of Illinois in Chicago. He said that look-alike and sound-alike errors are name confusion errors that often result in patient injury because of confusion in the process of ordering, dispensing or administering medications. He said there is not good documentation on how often errors of this type occur, although medication error reporting systems assert that about 20 to 25 percent of reported errors result from look-alike and sound-alike names. Cognitive, psychological, environmental and workplace factors contribute to these errors and it is the interaction between the cognitive and environmental workplace factors that is the complex cause of these errors. He suggests as a prevention strategy to identify which psychological processes are involved in each type of error. He said that the most comon kind of medication error occurs because of illegibility of physician handwriting where prescriptions are misunderstood by either pharmacists or nurses. He said that auditory misperception errors occur, as well, when oral orders are given. In addition, there are action errors, such as, when a data processor enters a name incorrectly. The probability of these types of errors increase when drug names look and sound alike. He recommends that in order to minimize errors due to look and sound-alike names, the following prevention strategies should be considered:

1. Don't use phonologically similar names,
2. Use short names instead of long names,
3. Maintain a quiet environment while dispensing or administering drugs.
4. Do not store drugs in alphabetical order
5. Eliminate handwritten and especially handwritten faxed prescriptions
6. Use barcodes, when necessary including adding the dose and indication on new drugs

Dr. Lambert has developed an automated system to measure phonological and orthographic similarity. By developing automated measures, he said, we may be able to study, in more detail, the relationship between similarity of drug names and the probability of error.

New Concepts for Error Proofing Pharmaceutical Trademarks

The last speaker was Mr. George DiDomizio, who is president of Gemini Trademark Services, Director of the Institute for Safe Medication Practices and is Vice President of Medical Error Recognition and Revision Strategies Inc. Mr. DiDomizio described the tools and techniques for error recognition based on visual and phonetic similarities of trademarks and demonstrated the concept of creative revision of flawed trademarks so as to lessen or eliminate error potential. He described the Error Recognition and Revision Strategy (E.R.R.S.) model which is a practitioner based model that relies on input and expertise of pharmacists, physicians, nurses and other health care specialists. Candidate trademarks are evaluated on visual and phonetic dimensions. The E.R.R.S. Model creates a simulation of the drug delivery process by using doctor written expressions of the candidate trademarks. It also uses voice prompts to simulate the verbal orders often used in a hospital or retail pharmacy setting. A variety of communication technology is used to interact with the participating practitioners, including the Internet, fax, computer and phone. For the visual evaluation portion of the process, input from practitioners deals with similarities to medical terms and abbreviations. Experience has shown, that in some cases, pharmaceutical trademarks can get entangled with medical orders when the product name and the medical terms are confusingly similar. Practitioner input also deals with similarities with names of existing brands or generic name products. Once the visual similarity data is assembled, the analysis is conducted in a clinical context, that is, the identified similarities are evaluated to determine the likelihood of confusion and a determination is made on the probability and extent of patient harm. In addition to the practitioner input, the E.R.R.S. Model makes use of supplemental data, including printouts from a computer program called PREDICT that combines pyscholinguistic theory and mathematical models to quantitatively measure the spatial and perceptual distance between various trademarks. This quantitative data is used to confirm the qualitative input from practitioners and to identify similarities that may have been overlooked by the practitioners. All data is collected and analyzed by the Institute for Safe Medication Practices and a report is issued to the client. Evaluations are made on the basis of a 5 point scale with 1 being assigned to trademark candidates with serious flaws and 5 being assigned to trademark candidates that are free of serious flaws.

Public Expert Panel

Panel Three was asked to discuss to what extent a user problem existed with FDA regulated products because of similarities in drug names, similar labeling, design and packaging of human drug, biologics, blood and blood products, vaccines and medical devices. They were also asked to consider what FDA's role should be in helping to minimize these types of errors and how effective collaboration could play a role in minimizing user errors. The following groups were represented: Alan Goldhammer, Ph.D. (Biotechnology Industry Organization), Joseph Cranston, Ph.D., (American Medical Association), Samuel Kidder, Pharm.D. (Health Care Financing Administration), Linda Golodner (National Consumers League), Paul Minicozzi, Ph.D. (Pharmaceutical Research Manufacturers Association), Thomas Granatir (American Hospital Association), Daniel J. O'Neal, M.A., R.N. (American Nurses Association), John Santell, M.S. (American Society of Health Systems Pharmacists) and Kenneth Kopesky (Health Industry Manufacturers Association). Dr. Bruce Burlington, who is the Director of FDA's Center for Devices and Radiological Health moderated the panel. Dr. Burlington began the discussion and said that there was a hypothesis presented at the meeting that there are millions of patients at risk, millions of errors occurring in medical practice delivery over a year's time and billions of dollars being expended to resolve the problems associated with user error. He asked the panel members to publicly debate the scope of the problem and to also consider who should take responsibility for resolving the problem.

Dr. Goldhammer said that the number of biotechnology products currently on the market is small and most if not all, are either injectable or parenteral products and usually administered by a health care provider. They are not usually the type of products would be dispensed from a outpatient or hospital pharmacy. He said that FDA, regulated industry and patients are all stakeholders in this issue. He said that the situation is complex, given the diversity of products and settings and that we still cannot truly define the scope of the problem. He agreed that it's a problem in the aggregate and said that no one group can attack the problem alone.

Dr. Minicozzi and said that there was a need to collaborate. He referred to a committee, which included representatives from AMA, APhA, American Society of Anesthesiologists, ISMP and regulated industry, who were charged with how to reduce errors related to confusing drug labeling. The committee made recommendations relating to revising the label and suggested, among other things, that there should be a standardized label format to reduce the incidence of user errors. He also referred to the FDA Modernization Act of 1997, and said that the legislation included many recommendations that came from the committee, which proved that collaboration could be an effective mechanism to address these issues.

Mr. Kopesky said that the device industry routinely interacted with CDRH staff and discussed issues like labeling, warnings and contraindications during the PMA review process. He said it would be helpful to have a collaborative effort with FDA, the industry and the health care professionals in order to successfully tackle these issues.

Dr. Burlington asked John Santell for the health professional perspective. Mr. Santell said that there is a collaborative role for government, industry, practitioners and consumers. He said that ASHP considered user error with FDA regulated products to be a significant problem and that FDA should take a stronger role in requiring manufacturers to provide evidence that they've had labeling and packaging review by practitioners and others who work in the medical field prior to approval of the product. Dan O'Neal said that in terms of measuring error, we significantly focus on medication administration and from the nurses point of view, a tremendous number of events occur with exponential impact at the point of the final user. He suggests that we should also take into consideration, not just the traditional setting of patient care, but increasingly the home and community based settings as well. Dr. Cranston said that since we were discussing error prone aspects of FDA regulated products, that FDA and industry should be most actively involved. Linda Golodner said that if consumers are supposed to be responsible for their own health, they need to know what's happening to them, what medical product is being administered and associated side effects, and that health care professionals should relay this information to patients as plainly as possible. She said there should be mandatory reporting and that reporting should be facilitated for consumers who are receiving care in a home care or outpatient clinic setting.

Tom Granatir said that his organization was unsure exactly how prevalent the problem of medical product errors are in health care, but they probably occur more often than originally thought and that medical product errors occur fairly frequently in hospitals. He said the problem of errors in medicine is shared by everyone, which speaks for the need for government action. He said FDA could help identify the nature of the problem and that the safety evaluation of drugs and devices is insufficient, if it doesn't include a summary of how the product is actually used in practice. Sam Kidder said that the HCFA position comes from a different perspective. He said that HCFA defines an error as a discrepancy between what product was actually administered to the patient based on the last signed orders of the physician. He said they would prefer that the administrator of the facility take responsibility, so proper resources can be applied in order to reduce the error rate. He said he thought in order to be successful, the effort would require a collaboration between FDA and industry.

Dr. Burlington said that there is tension between taking responsibility for one's actions versus implementing a no fault reporting system. Dr. Burlington asked the panelists to comment on this concept. Ms. Golodner said that there needs to be a central reporting mechanism and that the information should be shared with other hospitals and institutions. Dr. Burlington asked the panelists to elaborate on the perception that there is a serious problem in terms of how frequently errors occur in complex settings and said that often, we don't know how best to intervene or how to apply good intervention models. He asked the panelists to comment on whether FDA should have a broader role in reducing the incidence of confusion of product names. Dr. Minicozzi said that the marketing department at Wyeth Laboratories already reviews the labeling and names of drug products closely and conducts focus groups which include pharmacists, nurses and others to help eliminate confusion in drug labeling or naming prior to approval of the drug. Dr. Goldhammer suggested the need to explore a more standardized label or bar coding products prior to marketing as useful tools in minimizing medical product errors. Ken Kopesky said that standardized labeling in the medical device field would be difficult because of the variety of devices on the market. He said that physicians sometimes interpret device labeling differently and said that standardization would require collaboration among the stakeholders.

Dr. Burlington asked the health professional representatives whether now was the time for government to take a more proactive role in this area. Dr. Cranston said that he thought that the companies should consider implementing a qualitative type of analysis as described in Mike Cohen's talk (Failure Mode and Effects Analysis) prior to submission of product information and that FDA should more proactively consider this information as reviewers evaluate medical products for safety and efficacy. Mr. O'Neal said that better user training was important. Linda Golodner said that consumers would be appalled to realize that the Government is not more proactively involved. Dr. Burlington finished up the panel discussion by asking the panel whether other health care systems are managing the problem of errors with medical products better. Tom Granatir said that he thought the problems of error are fundamental to clinical practice, and will be the same in every health system.

Questions and Answers

Mike Cohen, from the Institute of Safe Medication Practices, said that perhaps 500 or 600 deaths a year could be attributed to labeling, packaging and naming issues and said that it would be unfortunate, if a few hundred thousand dollars couldn't be invested to test the products in an effort to prevent those deaths. Dr. Lambert, from the University of Illinois in Chicago, said that the low base rate of errors provided significant obstacles to making progress because it meant that very few systems, institutions or individuals would have incentive to make positive changes. Ken Kotenberg, from Hoechst, said that industry routinely searches trademarks, including trademarks that are registered and with trademarks in use. He said that the process includes identifying a trademark and registering it in some 100 different countries. Each country has an individual system making it a complex process. He said it is in industry's best interest to ensure that drug names are dissimilar and that industry tries to avoid name similarities to the greatest extent possible. He urged FDA to remember that FDA action in this area may ultimately result in application of standards that could lead to 20 or 30 trademarks (for one product) worldwide instead of having just one recognized trademark.

Open Public Hearing

Dr. Stuart Nightingale, who is the Associate Commissioner for Health Affairs at FDA, facilitated this session. He introduced the first speaker, Dr. Dean Gubler, who is a board certified practicing general surgeon with the Navy. Dr. Gubler said a more streamlined system is needed to report medical incidents that does not sacrifice report accuracy and detail. He also said that the primary objective for investigating errors should be to determine their root causes so that interventions can be designed that would discourage the repeat of similar errors. He said by encouraging voluntary, anonymous reporting from health care providers, large amounts of case-specific information can be rapidly screened and categorized to identify process and individual performance factors that contribute to errors with medical products.

Bruce McWhinney, who is Senior Vice President at Cardinal Health, commented on drug therapy complications, which he defined as more than medication errors, but also to mean therapeutic failures of all kinds, whether they are due to inappropriate dose, inappropriate drug or other causes. He said a growing body of research indicates that drug therapy complications can be significantly reduced through better use of pharmacists especially in acute care treatment teams and in ambulatory services. He said that in order to reduce the incidence of these errors, clinicians need to have better access to an expanded information system that incorporates a wider range of data and that information systems should be integrated to allow better measurements of costs, quality and outcomes across the spectrum of care. The addition of diagnosis and treatment algorithms to formularies could greatly reduce the risk of prescriber error, by providing physicians with up to date information at the time of therapeutic decision making. Finally, new ways of packaging drug products, and new reminder systems for patients and pharmacists, have been introduced to help increase compliance. Research has shown that the use of blister packaging, for instance, can dramatically improve compliance while cutting pharmacy costs.

Dr. Joshua Schwartz spoke next from the Centers for Disease Control and Prevention. He commented on the Vaccine Identification Standards Initiative (VISI) and highlighted some current problem-causing trends in the immunization program. He said accurate vaccination records are critical to the public health assessment of vaccine coverage, vaccine efficacy and vaccine safety. In one study evaluating the accuracy of computer immunization records transcribed from handwritten records, an overall transcription rate of error of at least 10.2 percent was found. To address the problem, CDC is hosting VISA in order to develop accurate, efficient, and more user friendly methods for labeling and transferring vaccine identifying information. VISA would like to establish common practices/formats for several innovations and improvements, for which prototype examples have been proposed including:

1. Bar coding vaccine packages to ease transfer of relevant information from vaccine vials/packages to patient medical, billing and inventory records and into immunization registries using electronic scanning technologies.
2. The inclusion of vaccine packaging peel-off sticky labels in plain text and barcodes for placement into medical and other records.
3. Rethinking and standardizing lot numbers, potentially incorporating digits to identify manufacturer and vaccine type to reduce errors.
4. Developing standards to clarify essential vaccine information on vaccine packaging.

Dr. David Classen, who is a practicing infectious disease physician, at LDS Hospital in Salt Lake City, Utah, presented data from a large study where 202,000 patients were studied prospectively by computer through the occurrence of adverse drug events and medication errors. He said that voluntary reporting of adverse drug reactions was not effective and said, the real need in detecting adverse drug events is automated algorithms that contain a significant amount of information in the computerized record. According to Dr. Classen's study, out of 202,000 patients studied, there were 138 instances, where there was overlap between medication errors and adverse drug events. He said that over 20 hospital organizations have said that they have reduced their medication errors by 50 percent, but are not showing any impact on patient outcomes. Dr. Classen said that one could prevent 95 percent of all medication errors in a specific hospital, but still not show any impact on patient outcomes.

He said in order to develop successful strategies to prevent adverse drug events, you must focus on the events and not those things that might lead to adverse events, but often do not. Secondly, it is necessary to assemble a multi-disciplinary team of doctors, pharmacists and nurses who can act as a liaison team and still be recognized as primary care givers. Third, he suggests that we don't use the term "error", because the term is threatening to health care professionals. These strategies were evaluated in order to see if they reduced the number of adverse drug events and found that early notification of adverse drug events prevented them from moving onto severe adverse drug events, but did not actually impact on the numbers of adverse drug events. Standardizing drug administration rates, allergy tracing and drug interaction tracking cut the adverse drug event rate by about 5 percent. Dosage adjustment of renal function, adjusting doses of drugs that are excreted by the kidneys for the patient's age, weight and renal function cut adverse drug events by about 35 percent, but the most effective strategy for preventing adverse drug events and demonstrated a 75 percent decrease in adverse drug events was a comprehensive disease management program. Dr. Classen said that the program was effective because when one focuses on adverse drug events and errors, there is only a slight impact on all potential drug related problems, and only an overall comprehensive disease management program can impact all adverse drug events.

Dr. DeRichemond, who is Associate Director of the Health Devices Group at ECRI was the next speaker. He said that there is more of a training burden with regard to utilizing medical devices now in health care, because there are more generalists performing duties held by only specialists a few years ago. He said that the need for engineering solutions to medical device problems is also increasing. To provide solutions, engineers need to understand the current problems associated with medical devices. He said FDA needs to help engineers identify and learn from mistakes and quickly implement corrective solutions.

The next speaker was Mr. William Ellis, who spoke on behalf of the National Coordinating Council for Medication Error Reporting and Prevention. Mr. Ellis highlighted a few of the Council's recommendations including:

1. FDA should restrict the use of any printing on the cap or ferrule of injectable drugs except to convey warnings. Innovative labeling should be developed to aid practitioners in distinguishing between products of similar names.
2. FDA should discourage industry from printing company logos and company names that are larger than the type size of the drug name. The Council supports the USP/FDA advisory panel on simplification of injection labeling and recommends that this effort be expanded to areas like package inserts and dosage form labeling.
3. The Council encourages collaboration among regulators and standard setters with industry and health care professionals.
4. The Council encourages the use of innovative packaging and labeling techniques such as tactile cues to help practitioners identify differences among critical drugs.
5. The Council urges regulators and standard setting organizations to encourage the printing of drug names, both brand name and generic, and the strengths of both sides of injectable and IV bags and over wraps. For large volume parenterals and IV piggybacks, the name of the drug should be readable in both the upright and inverted positions.
6. The Council urges manufacturers to expand the use of bar coding for new drug products.

Eleanor Vogt, who is a senior fellow from the National Patient Safety Foundation spoke unofficially for the Foundation. She had three recommendations for the FDA including to:

1. Seriously explore the need to broaden the FDA approach to error by looking at the costs to patients and to the system where there is no or erratic follow through.
2. Continue to encourage compliance packaging, PPIs and easier dosing patterns.
3. Encourage and challenge practitioners and manufacturers to apply systems thinking and human factors analysis to the broader area of risk which is patient follow through.

Dr. Neil Lerner, who managers the human factors program for Westat was the last speaker during the open public hearing. He said that problems related to warnings, labels, instructions, symbology alarms, hazard recognition, risk taking behaviors and errors in product use have been explored and that he is struck by the number of common issues and themes present. He said that if we are interested in finding innovative solutions to these problems, we should strive to find solutions to common types of problems. He said he thought there was the possibility for shared efforts and economies of scale that might allow us to treat these problems differently. For example, there are already hospital data systems and insurance industry cooperative studies in place within the automobile or product safety areas. He said they have learned that labels can serve many functions, but that they really only work if one begins with a user-centered approach. He said that some other agencies are searching for alternatives to regulation and standard setting and they are developing very specific kinds of shared evaluation procedures, guidelines and safety audit methods that are more flexible and better able to respond to changing products and technologies.

Closing Remarks

Ms. Smith-Holston said that we had heard many ideas that were both provocative and stimulating at the workshop. She acknowledged the fact that minimizing medical product errors is a very complex issue with many factors that contribute to the problem, some of which are mechanical, human, social and environmental. She said that we all have a role to play as we work toward solutions to minimizing medical product errors. She closed the workshop by saying that FDA will be assimilating the information presented at the workshop and considering the appropriate next steps to minimize the occurrence of errors with medical products under the FDA's jurisdiction.