AmerisourceBergen Bridge Medical Features Resources
Bridge Medical

To improve clinical practice and quality of care by providing a learning opportunity that enhances the participant's understanding of a systems approach to medication errors and error prevention.


When you have completed this course, you will be able to:
  Identify the situations, circumstances and actions that contributed to a fatal medication error in one hospital.
  Identify systems, processes, and conditions that could contribute to errors in any clinical environment.

It was a tragedy that should never have happened. A newborn infant was given a tenfold overdose of the intramuscular (IM) medication penicillin G benzathine by intravenous (IV) route that resulted in his death. The parents were devastated at the loss of their child; the hospital staff was devastated at the loss of their patient. But the devastation did not stop there. As a result of the medication error and subsequent death, three nurses were indicted on charges of negligent homicide█ the first time that nurses had faced criminal charges for a medication error in this country.3 In January 1998, two of the nurses plead guilty in response to a plea bargain; the third nurse, who refused the plea bargain, was acquitted at trial.5

When an error as traumatic as this occurs, it is only natural to want to find the person "who did it," assign them responsibility, and impose some punishment. With the finger-pointing approach, the most proximal cause of the error is generally the one that bears the burden of blame. Unfortunately, this approach does not take into account that most medication errors occur, not as a result of a single act by a single individual, but as a result of multiple, cascading events compounding one another until a catastrophe happens. Most errors occur as a result of "a chain of events set in motion by faulty system design that either induces errors or makes them difficult to detect," rather than lack of care and concern on the part of our caregivers.2

A Systems Approach

A system can be defined as "an interdependent group of items, people, or processes with a common purpose."2 The systems approach to error management has been a fact of life in aviation and nuclear power management for many years. When an air travel incident occurs, the first question is not who caused the accident, but what caused the accident. In medicine, the systems approach has been slow to take hold. Perhaps it is because we don't want to believe that physicians, nurses, and pharmacists are fallible. As professionals, we are trained to believe that perfect performance is not only expected, it is also achievable. That if we know enough, take enough care, and try hard enough, our patients will be safe and protected. As Lucian Leape, author of "Errors in Medicine" noted, the result is that we come to view error as a failure of character.1

Psychologists would call this a false belief because the reality is that no matter how knowledgeable, no matter how careful, no matter how hard we try, human beings make errors! Systems that rely on error-free performance are doomed to failure. Once we recognize that fact, we have a much more powerful mechanism for discovering the real cause of catastrophic errors and preventing them in the future.

Errors and Human Cognition

Cognitive psychologists and human-factors specialists have been studying how and why people make errors for many years. Cognitive psychology has to do with how human beings think. Human-factors specialists deal with the man-machine interface in complex environments.1 Their work in airplane cockpits and in nuclear power plant control rooms can serve as a helpful model to healthcare in redesigning systems to reduce errors. Very simply, they have found that there are two modes of mental functioning: automatic and problem solving.

Problem Solving?

In the automatic mode, mental processes are fast and effortless. We really don't have to "think" in this mode, our actions are unconscious and require our attention only if there is a change. This automatic mode is possible because human beings are able to carry a vast array of mental models that are expert and fine-tuned processes for some minute recurrent aspect of our world.1 Once we have a task "down," it becomes a part of automatic mode thinking.

The problem-solving mode, on the other hand, may require intense mental concentration. We have to gather information and compare it to stored knowledge, or apply some decision rule, in order to know how to act. As a result, problem-solving processes are much slower, sequential, effortful and difficult to sustain.1

The errors that occur when we are functioning in these two mental modes are also different. When we are functioning in automatic mode, our errors are called "slips" in that they usually result from distractions or failure to pay attention at critical moments. An example of an error in automatic mode would be walking into a room only to discover you cannot remember what you came in to do.

Errors that occur when we are functioning in problem-solving mode are thought of as "mistakes." These are rules-based errors that might occur when the wrong rule is chosen, either because we have misperceived the situation, or perhaps because we just misapplied a rule. Errors can also arise when we have gaps in our knowledge base, especially when we are confronted with an unfamiliar situation for which we have no programmed solution. Other factors that effect our functioning in problem-solving mode are pattern matching, biased memory, the availability heuristic, confirmation bias, and overconfidence.

Pattern matching is the tendency to find patterns in situations so we can apply previously thought out solutions. Biased memory is the tendency to over-generalize because familiar patterns are assumed to have universal applicability. The availability heuristic is the tendency to use the first information that comes to mind. Confirmation bias is the tendency to look for evidence that supports early working hypotheses and ignore new information that contradicts it. And overconfidence is the tendency to believe in the validity of the chosen course of action and to favor evidence that supports it.2 All of these factors effect our problem-solving ability.

Various physiological and psychological factors can contribute to inattention and distraction and can play a part in both automatic mode errors and problem-solving mode errors. Physiological factors include fatigue, sleep loss, alcohol, drugs, and illness. Psychological factors include distraction due to other activity ("busyness"), as well as emotional states such as boredom, anger, fear, and anxiety. Psychological factors can be triggered by external factors such as overwork, interpersonal relations, and other forms of stress. Environmental factors such as noise, heat, motion, and visual stimuli can also divert our attention and lead to slips and/or mistakes.1

While an appreciation for the psychological and physiological factors that contribute to errors is useful in understanding that errors are an essential part of being human, they provide limited insight into ways of avoiding or preventing errors. That is where a systems approach it most beneficial. We know that human beings will naturally have "slips" and make "mistakes." Systems research and the work of human-factors specialists inform us that certain aspects of our systems can either add to the likelihood of error occurring, or act as a fail-safe to prevent errors.

Errors can be termed active or latent. An active error is one whose effect is immediate. For instance, turning left against on-coming traffic has an immediate effect. Either you are hit broadside, or just avoid a collision. Latent errors are those situations or errors whose effects are delayed - the proverbial "accident waiting to happen."4 In our left turn example, perhaps the intersection is poorly controlled, or trees or a curve in the roadway impair the visibility. As a result, every time a driver needs to make a left turn, the potential for disaster is high. Accidents rarely result from a single error, latent or active.1

The primary objective of any system review is to identify those latent errors█ accidents waiting to happen, and redesign or alter the system to recognize and correct the error before it happens. If it is not possible to prevent an error from happening, the next best alternative is a system that self-corrects when an error is detected. And if a self-correcting system is impossible, systems should be designed to detect errors and surface them as soon as possible so remediation can start. This requires that systems be designed with feedback and monitoring mechanisms as an integral part of the system. In the case of critical mechanisms and processes, redundancy should be built in to compensate in the event the primary system fails.4

Effective accident and error prevention focuses on root causes - errors in the system design and implementation, not on the errors themselves or the individuals making them. In order to do that, certain principles should be followed. For instance, tasks should be simplified. "Forcing functions" should be integrated that make it impossible to act without meeting a precondition. And processes should be standardized to reduce variability and reinforce pattern matching.1

Proximal Causes
  Proximal Causes

Leape, et al, have identified a number of "proximal causes" of medication errors. Proximal causes are defined as "the apparent 'reason' the error was made."2 Proximal causes are broad categories where the underlying system problems that result in medication errors may be found.

By grouping proximal causes into broad categories, they become a useful means of focusing further inquiry, but they are not the true cause of errors.2

Facts of the Case
*All the facts of this case, as presented in this section, were taken entirely from Smetzer, Judy L. "Beyond Blaming Individuals: Lesson from Colorado," Nursing98 (May 1998).
The case that follows is not unlike many incidents involving medication errors. A number of factors contributed to the creation of a complex series of events. And while you may want to assign blame to one or more of the parties, try to suspend that impulse and look for the systems failures - the proximal causes instead.


As you proceed through this case, we will provide you with information about the incident cited at the beginning of this article. You will have an opportunity to give your opinion about the type of system failures that may have contributed to the medication error and compare your responses to those of all other individuals taking this course. At the end is a test. If you are taking this course for continuing education credit, click on the Exam button to complete the test and submit your answers.

Deciding to Treat

Miguel Sanchez was born on October 15, 1996, a healthy, 7 pound (3.2 kg) male infant. Shortly after his birth, the hospital staff became aware that his mother had a history of syphilis. It was unclear from the mother's prenatal record when she had had the syphilis and whether or not it had been successfully treated. Further complicating the situation was the fact that both parents spoke only Spanish, and while translators were available, the staff was still unable to determine whether the syphilis had been treated. This caused the staff to have some concerns about the possibility of congenital syphilis.

In addition to the inability to gather sufficient information about the mother's medical history, the language barrier also presented some difficulty in discussing treatment options. So, while there was no urgent clinical reason to treat Miguel immediately, the neonatologist was not convinced the parents understood the importance of follow-up. Therefore, the decision was made to investigate the possibility of congenital syphilis in little Miguel and treat him before discharge.

Congenital syphilis is rarely treated in the hospital and the staff was relatively unfamiliar with it. The neonatologist called a local infectious-disease specialist who recommended diagnostic studies including a lumbar puncture. The consultant also recommended a single dose of penicillin G benzathine, IM, at a dose of 50,000 units/kg. Unfortunately, the neonatologist did not document the recommendation he had received until the following day, after Miguel's death.

Another staff member called a health department epidemiologist who made the same recommendation. She then documented the recommendation on the physician order sheet in the space provided for progress notes. Unfortunately, her notation was inaccurate. The note stated the health department had recommended "penicillin G 50,000 units/kg" when in fact, the recommendation had been for "penicillin G benzathine." She did not record the recommended route of administration in her note. At the time of administration, this was the only documented recommendation.

Writing the Order

The day after Miguel was born, a different neonatologist performed the lumbar puncture and wrote an order for one dose of "Benzathine Pen G, 150,000 U IM". The physician's written order was unusual in that instead of writing "penicillin G benzathine," he wrote "Pen G" on one line and placed "Benzathine" on the line above "Pen G." Further confusing the order was the fact that the route, "IM" appeared to be written over and looked somewhat like "IV". Also, "units" was abbreviated as "U," which could be easily misread as one or two additional zeros.

Remember, at this point, the only documented recommendation was the one written by the staff member who called the health department, and that was inaccurate.

Order Processing

When the order arrived in the pharmacy, there was no pediatric pharmacist working. The pharmacist on duty was also unfamiliar with the treatment for congenital syphilis and she had little knowledge about this rarely used, non-formulary drug. She reviewed the health department's recommendation (inaccurately documented on the physician order in the progress note section) and Drug Facts and Comparisons to determine the infant dose. Unfortunately, she misread the dose in both sources as 500,000 units/kg rather than the correct infant dose of 50,000 units/kg. She also misread the physician order as 1,500,000 units instead of 150,000 units. Since she was under the mistaken understanding that the correct infant dose was 500,000 units/kg, it is easy to see how the pharmacist could have misinterpreted the "U" as an extra zero, thus seeing 1,500,000 units on the order.

When she entered the tenfold overdose into the pharmacy computer system, the system did not generate any warnings about exceeding the maximum dose for a neonate. As a result, the pharmacist prepared an incorrect dose: 1,500,000 units.

Drug Dispensing

Because the nursery did not have a unit dose system, the pharmacist prepared two pre-filled syringes of penicillin G benzathine, 1.2 million units/2 ml each. She put green stickers on the plungers reading "Note dosage strength" to indicate that the entire contents of the two syringes should not be administered.

The syringes were placed in a plastic bag, and a label was affixed with directions to administer 2.5 ml of the medication IM to equal a dose of 1,500,000 units (one full syringe and one-quarter of the other). The correct dose should have been 150,000 units/0.25 ml. No other warning labels were affixed to the drug to alert the nursing staff that penicillin G benzathine should be administered IM only.

Perhaps because she was not a pediatric pharmacist, the pharmacist was unaware that a maximum volume of only 0.5 ml can be safely administered to an infant per IM injection. If she had known that, the tenfold overdose might have been detected at this point since the dose as dispensed was going to require five separate injections.

After preparing the medication, the pharmacist noticed that the medication in one of the pre-filled syringes had expired. After replacing the syringe, another pharmacy staff member dispensed the drug without checking it against the original order. Once again, the tenfold overdose was missed.

Preparing for Administration

After noting the order for penicillin G benzathine, Miguel's primary care nurse and a neonatal nurse practitioner researched the treatment of congenital syphilis by consulting the 1994 Red Book: Report of the Committee on Infectious Diseases. One drug recommended to treat congenital syphilis was penicillin G benzathine IM. However, the source, which was not a specific drug reference, did not warn that the drug could be administered IM only.

When the medication was delivered to the nursery, the primary care nurse expressed concern to her colleagues about the five injections Miguel would need. If the drug had been dispensed correctly, just one IM injection of 0.25 ml (150,000 units) would have been needed. Hoping to prevent Miguel unnecessary pain, an advanced-level nursery RN and the neonatal nurse practitioner decided to investigate giving the drug IV instead of IM.

The two nurses consulted Neofax '95 to determine if penicillin G benzathine could be administered IV. The monograph did not specifically mention penicillin G benzathine. Instead, it identified the treatment for congenital syphilis as aqueous crystalline penicillin G by slow IV push or penicillin G procaine IM.

As a result of this lack of specific reference to penicillin G benzathine, and being unfamiliar with the various forms of penicillin G, the neonatal nurse practitioner thought that "Benzathine" was a brand name for penicillin G. This misconception was further reinforced by the unusual way the physician had written the order: with "Benzathine" on the line above "Pen G." And since the health department recommendation had been inaccurately recorded as just "penicillin G" with no specified route of administration, the misconception continued.

In addition, many sources use ambiguous synonyms when referring to different forms of penicillin. Believing that "aqueous crystalline penicillin G" and "penicillin G benzathine" were the same drug, the neonatal nurse practitioner concluded that the drug could be safely administered IV rather than IM.

Drug Administraion

While hospital policy did not clearly define the prescriptive authority of non-physicians, nurse practitioners in this hospital routinely prescribed medications within the scope of their practice. And in fact, routes of administration had been changed by nurse practitioners before. The neonatal nurse practitioner also thought she was acting under a national protocol that allows nurse practitioners to plan, direct, implement, and change drug therapy. Consequently, she decided to administer the drug IV.

While preparing to give the medication, neither nurse noticed that the syringes were labeled with a manufacturer's warning: "IM use only."

IM use onlyThe warning was not prominently placed and to view it, the nurse would have to rotate the syringe 180 degrees away from the drug name. Also, once the pre-filled syringe was assembled for use, the orange plunger concealed part of the "M" in "IM" making it possible to misread the warning as "IV use only." Once again, the tenfold overdose went undetected.

In addition to the packaging problems, the manufacturer's labeling on the syringes, as well as the pharmacy label, expressed the dose using multiple zeros: "1,200,000 units" and "1,500,000 units" rather than "1.2 million units" and "1.5 million units." Perhaps if the dose had been expressed thusly, the tenfold overdose would have been detected, but it was not.

Penicillin G benzathine is a white, milky substance. This might have caused the nurses to pause, but they were also aware that some IV substances, such as lipids and other lipid-based products, looked milky and still could be safely given IV. So, even though this conflicted with the widely recognized rule that only clear liquids should be given IV, the nurses did not recognize any problem with giving penicillin G benzathine IV.

And so, believing they were sparing the baby unnecessary pain, the nurses began to administer the first syringe of the drug by slow IV push. After about 1.8 ml of the medication had been administered, the infant became unresponsive. Resuscitation efforts were unsuccessful and Miguel died.

Systems Analysis

As we mentioned before, it is tempting to assign blame for this error to an individual, but it is also clear that in this case, many individuals were involved with many opportunities to detect the error. All the parties involved had to deal with the tragic results of their mistakes, and since Miguel's death resulted in criminal charges, the three nurses also had to deal with the additional trauma of a criminal trial.

In preparation for the trial, the staff at the Institute of Safe Medication Practices worked with the nurses' defense attorneys to perform an in-depth system analysis of the medication error and provide expert testimony at trial. As a result of that analysis, they identified over 50 different system failures that allowed the error to develop, remain undetected, and ultimately, reach this infant.5

Recall for a moment Leape, Bates, and Cullen's list of proximal causes of medication errors. When we review the major system failures in the Miguel Sanchez case, most of the errors would fall into one of these categories. It is clear that if even one of these failures had not occurred, the error would have been detected and the baby would not have been harmed.

The systems approach can be a powerful framework in correcting existing system failures and in identifying and remediating potential system failures. It is up to us, the care providers, to stop blaming individuals for errors and start looking at the root causes within our systems that are the true culprits.