Celebrated Accidents

The "celebrated" cases are medical accidents that have attracted a great deal of attention from the public and the press (Table 1; Figure 1). The reports of these cases have led to a variety of reactions from health care professionals, regulators, and the public. Interestingly, many of these cases have achieved such a level of prominence in the collective public psyche that one can elicit a collection of images simply by mentioning the "Florida wrong leg or Willie King case" or the "Betsy Lehman case" or the "Libby Zion case." All of these cases evoke our empathy for some tragic loss.

These cases also have become symbolic in other ways. The case of Willie King in Florida, in becoming the "wrong leg case," captures our collective dread of wrong site surgery. The death of Libby Zion has come to represent not just the danger of drug-drug interaction but also the issues of work hours and supervision of residents - capturing symbolically our fear of medical care at the hands of overworked, tired, or novice practitioners without adequate supervision. Celebrated cases such as these serve as markers in the discussion of the health care system and patient safety. As such, the reactions to these tragic losses become the starting point for discussions of obstacles and opportunities to enhance safety.

The sourcebook distributed to participants contained background on a selection of these cases. Most of the available material comes from newspaper articles on a specific case. Also included were two broader perspectives from reporters looking across multiple health care accidents and attempting to synthesize a more coherent picture of accidents in general (Lisa Belkin, New York Times, June 15, 1997; Steve Twedt, Pittsburgh Post-Gazette, October 24-31, 1993).

The sourcebook also included a column by the news commentator Sidney Zion, father of Libby Zion. This commentary appeared following a media briefing conducted by the National Patient Safety Foundation. It conveys some of the charged atmosphere surrounding public discourse on patient safety issues. It may presage the sorts of communication difficulties that will confront those who seek to develop and explore a more technically grounded view.

Typically there are no independent investigations of the sequence of events and contributors to the outcome such as those done by the National Transportation Safety Board (NTSB) following aviation accidents. As a result, we often must rely on news reports, but other sources of information may be available. In the "wrong leg" case there is also an official document prepared by a Florida hearing officer. This provides the legal rationale for the revocation of the surgeon's license and gives an account of the chain of events leading to the amputation of the wrong leg. The materials for the Ben Kolb case include a set of statements giving the perspectives of the hospital CEO, the risk manager, the physician, the medical liability insurer, and the family.
 

The View of Patient Safety from Celebrated Cases

The First Story
Each of the celebrated cases consists of a relatively straightforward, simple, and easily understood story constructed after the accident. This first story typically explains the accident in terms of a simple cause, usually human operator error. The people held responsible are those closest in time and space to the final outcome, especially those who, it is believed, could have acted in another way that would have led to a different outcome. Their behavior, with knowledge in hindsight of the tragic outcome, appears to be outrageous, willful disregard of critical cues or factors. In retrospect, each failure seems preventable by relatively simple means, such as new policies and procedures or calls to increase the "vigilance" of practitioners. Finding the culprit ends the investigation.

Each celebrated case has come to represent a specific threat to safety for the public and for people in health care. Figure 1 shows schematically the relationship between the celebrated cases and the sources of failure they seem to represent. However, the reactions in these celebrated cases provides only a partial view of the hazards and of the factors that contribute to failures.

In the "wrong leg case," for example, public attention focused on the surgeon. However, this case represents a larger class of failures that arise because of the symmetry of the human body. Bilateral symmetry (paired organs, limbs, etc.) creates the inherent risk of wrong site, wrong side, wrong limb failures. This risk is well known. Health care practitioners and organizations recognize this risk and have a variety of defenses against just this sort of accident. Cases like the Florida wrong leg amputation are situations where all the defenses broke down or were ineffective. They point, not so much to inadequate defenses, as to a systemic inability to maintain these defenses in working order in the face of a variety of pressures. Indeed, the accident is a potential source of data about the larger system that delivers care, not simply about a flawed individual. Understanding the factors that bypassed or undermined the defenses in this case would help us learn in ways that could be applied to other situations, practitioners, and organizations.

The response to failure is the most significant feature of each of the celebrated cases. The affected organization usually creates new policies or procedures in the hope of forestalling any repeat of this particular accident. For example, several celebrated cases have involved medication misadministrations. Responses have included efforts to more tightly control the use of the particular drugs by imposing the requirements for more elaborate checks and additional steps in prescribing and dispensing medicines.

However, some drug misadministrations point to a broader risk related to drugs with a low therapeutic index. These are drugs where the effective dose is near the toxic dose. Recognizing this class of drugs and associated risks can help illuminate weaknesses in the defenses deployed to mitigate those risks. Drug-drug interactions, e.g. the Libby Zion case, represent a hidden low-therapeutic index situation that contributes to the outcome in complex ways.

The response to the failure also provides significant data about how attributions of causality are made and how responses to a widely publicized failure are driven by a variety of factors. The limited ability of the regulatory bodies to influence safety in a direct way is one remarkable characteristic of the 'wrong leg' accident. The regulators' choice to "send a message" by revoking the surgeon's medical license is itself an interesting feature of the case. The decision to send such a message and phrase it in this fashion signals a set of beliefs about which factors lead to failure, which interventions can change those factors, and how health care practitioners are expected to react to these messages. These beliefs constitute one model of how people contribute to safety and risk. The scientific question is, do these beliefs correspond to accurate models of the factors that affect success and failure and of the factors that enhance or degrade the performance of health care practitioners?

Taken together, the celebrated cases were tantalizing to the researchers. They indicate the potential for catastrophic failure in health care. They also demonstrate the way that these failures can capture public attention, evoke outrage, and provide impetus for regulatory action.

But the cases are also remarkable for their limitations. The workshop participants rapidly tried to move beyond this "first" story into a discussion of the deeper "second" story that lies behind such cases. However, in each instance, the story told after the event is too simple. Its details are too limited to serve as the basis for understanding the interplay of the multiple contributors that led to the accident. While the accounts of the celebrated failures do tell us a great deal about the social response to failure, the participants observed that the celebrated cases do little to broaden our understanding of the other risks that exist in health care, the sources of these vulnerabilities, or the means for reducing them.

To better understand systemic vulnerabilities to failure and to see how failure is usually prevented requires collecting a different kind of data, analyzed in different ways, that reveal a different story. To accomplish this requires analysis based on concepts grounded in the research base about the factors that affect the many different kinds of human performance relevant to health care settings. In particular, the researchers recognized the impact of hindsight bias on the construction of these first stories of accidents.

The Hindsight Bias
The tendency to attribute accidents in health care to simple causes such as isolated human failures is derived in part from a particular form of bias that clouds post-accident reviews of human performance. It is well documented that knowledge of outcome biases our later judgments about the processes that led up to that outcome (See Figure 2). The way we look back is shaped by the outcome. That outcome knowledge, however, was not available to the participants before the fact. In looking back we tend to oversimplify the situation the actual practitioners faced. This blocks our ability to see the more complicated, richer story behind the label human error.

The hindsight bias effect is a well reproduced research finding relevant to accident analysis and reactions to failure. In the typical study, two groups of judges are told a story and asked to evaluate the performance of characters in the story. The story is identically told for each group of judges with a single exception: the difference is the outcome of the story. One group is told the episode ended in a poor outcome (death, significant loss, etc.). The other group is told that the outcome was good (minor injury, insignificant loss or even some gain). The two groups of judges consistently differ in their assessment of the story characters' performance. Judges told of the bad outcome assess the performance as flawed. Judges told that the outcome was successful assess the performance as acceptable. The differences are stark, repeatable, and strong. In fact, hindsight bias impacts judgments even when judges are warned that the outcome knowledge may influence their ability to make assessments.

It is clear that hindsight bias poses a great obstacle to understanding patient safety through celebrated cases. The powerful outcomes of these accidents shape the way that post-accident (looking back) judgments of human performance are made. This bias limits the value of these cases because the debris of outcome obscures the complexity of the situation confronting practitioners. This leads to simple "first stories" of accidents and, paradoxically, limits what can be learned about safety from such events.

Uncelebrated Cases

The Second Story
The next phase of the workshop provided a contrast to the celebrated cases in the form of "uncelebrated" cases where some research base was available. The cases highlight some of the factors that affect success and failure in the practice of medicine. More importantly, the cases demonstrate the kinds of factors that affect the success and failure of research on human expertise and its role in system performance. In contrast to the celebrated cases, a multi-faceted story about how the system works and how it sometimes fails unfolds in the investigations.

In this section, we provide a summary of what was presented about the three uncelebrated cases. Then we draw out some of the larger implications of each case for patient safety in general based on the discussion at the workshop and the research results themselves.

While the details of the uncelebrated cases are distinct, they have at least five features in common.

First, each uncelebrated case shows that bad outcomes flow not from single-point failures but from a set of factors. The research reveals that these factors are each necessary but only jointly sufficient to cause an accident. The analysis in the uncelebrated cases exposes the system issues and latent factors that contribute to failure.

Second, these investigations show that enhancing safety begins with efforts to understand not just the sources of failure but also the sources of success. System operations are seldom trouble-free. In every close examination of complex systems in operation, observers find many more opportunities for failure than actual accidents. The difference between the high potential for failure and the low rate of failure is produced largely by practitioners.

Much of expertise and skill is directed towards preventing poor outcomes or recovering from problems before their consequences impact on the patient. Each investigation shows how practitioners resolve conflicts, anticipate hazards, accommodate variation and change, cope with surprise, work around obstacles, close gaps between plans and real situations, detect and recover from miscommunications and misassessments. In these activities practitioners regularly forestall or deflect potential accident trajectories.

Put another way, human practitioners are not so much the cause of occasional sporadic accidents as they are the active agents that regularly contribute to success. When they carry out their roles successfully, they are the active creators of safety. Safety research tries to identify factors that undermine practitioners' ability to do this successfully.

Third, the research results shift attention away from the people closest to the accident and toward the blunt end of the system where regulatory, administrative, and organizational factors reside. Complex systems such as health care or aviation have both a sharp end and a blunt end (Figure 3). The sharp end is where practitioners interact directly with the hazardous process in their roles as pilots, mechanics, air traffic controllers, and, in medicine, as nurses, physicians, technicians, pharmacists and others. At the blunt end of the health care system are regulators, administrators, economic policy makers, and technology suppliers. The blunt end of the system is the source of the resources and constraints that form the environment in which practitioners work. The blunt end is also the source of demands for production that sharp end practitioners must meet. The demands are often conflicted, as when the blunt end provides incentives for greater production while simultaneously demanding lower rates of failure.

The more safety researchers have looked at the sharp end, the more they have realized that the real story behind accidents depends on the way that resources, constraints, incentives, and demands produced by the blunt end shape the environment and influence the behavior of the people at the sharp end (Reason, 1997). Detailed examination of accidents in these systems consistently shows that the ability of sharp end practitioners to defend against failure in these cases depended directly and indirectly on a host of blunt end factors rather than on the isolated "error" of human practitioners.

Fourth, the research methods described in reviewing the uncelebrated cases demonstrate, in part, of how research that ultimately improves safety is done. The combination of methods chronicled in these cases may be unfamiliar to many, but they represent the kinds of techniques that have been developed to understand the role of human performance, human-machine cooperation, and cooperative work in complex evolving situations.

Fifth, and perhaps most important, the research on the uncelebrated cases points to areas where substantial progress can be made. Significantly, these are not single, local fixes or "magic bullets." Rather, the research reveals a set of factors involved in failure and shows that there are multiple directions for improvements that need to be coordinated in order to make progress on safety.

 
Uncelebrated Case #1:
Bile duct injuries during laparoscopic cholecystectomy

Uncelebrated Case #2:
Antibody misidentification and transfusion reactions

Uncelebrated Case #3:
Drug misadministrations via computerized infusion devices in the operating room

Day One Footnotes

See Woods et al. (1994), chapter 6 for an overview of the research. For the original studies see Fischhoff, B. (1975). Hindsight ? foresight: The effect of outcome knowledge on judgement under uncertainty. Journal of Experimental Psychology: Human Perception and Performance, I, 288-299 and Baron, J. and Hershey, J. (1988). Outcome bias in decision evaluation. Journal of Personality and Social Psychology, 54, 569-579. For a replication in medicine, see Caplan, R., Posner, K., and Cheney, F. (1991). Effect of outcome on physician judgements of appropriateness of care. Journal of the American Medical Association, 165, 1957-1960.

Return to document