The Wrong Patient Free

Of all of the errors we make in delivering health care, this case surely represents one of the most disturbing. Despite occasional local news coverage of such adverse events (1-3), few data are available to document their incidence. The Joint Commission on Accreditation of Healthcare Organizations maintains a national database of “sentinel events” (4, 5), which include errors such as this one. Reporting to the Joint Commission is voluntary; the database contains 17 reports of an invasive procedure done on the wrong patient over the past 7 years (Schyve P. Personal communication, 31 January 2002). Additional information about adverse events is compiled by individual states, at least 15 of which maintain their own error-reporting systems (6). New York State has a long-standing and recently revised mandatory reporting system; it has received reports of 27 “incorrect patient/invasive procedure” incidents from April 1998 through December 2001 (Heigel F. Personal communication, 14 February 2002).

The marked disparity in the number of events chronicled by these two databases—one voluntary, one mandatory—suggests that the voluntary Joint Commission database is incomplete (5). But even mandatory state reporting systems may underestimate the true incidence of “wrong-patient” procedures. All error-reporting systems depend on hospitals' internal incident reports as sources for their data, and research has shown that clinicians file incident reports for only a small percentage of actual errors (7-9). A recent analysis of the New York system, for example, determined that for one of the adverse events for which reporting is required (deaths within 48 hours of surgery), only 16% of cases were reported in 1999 (10, 11).

The medical literature is largely silent about this problem. We found a small number of studies showing errors of patient misidentification in the transfusion of blood products (12), injection of radionuclide material (13), and administration of chemotherapy (14). We could not find a single study or case report on the problem of wrong-patient invasive procedures. Although the New York data may provide a lower bound frequency estimate, given the dearth of research and the limitations of error-reporting systems, we conclude that we do not know how often this type of event occurs. However rarely these events occur, all health care delivery systems should strive to eliminate them entirely.

Comment from Ms. Morris upon being awakened by RN₂: I was sound asleep, and a lady comes in and says, “Oh, you have an appointment in room so-and-so.” I said, “No, I don't. ”I told my children and I told the doctors I did not want to do anything else right then … She just zoomed in and took me on out of there.

How could Ms. Morris, a native English speaker and a high school graduate, have signed a consent form for a procedure she knew she was not supposed to undergo—a consent form that indicated her agreement to possible cardiac surgery to implant a defibrillator? In theory, the process of informed consent should protect both patients and caregivers from adverse events such as this by providing patients the information they need to become full participants in decisions about their care. In practice, however, the process of obtaining informed consent is often deeply flawed. Obtaining consent is frequently delegated to an overburdened or exhausted physician who has not met the patient previously and does not know the details of the medical history. Cultural or social barriers to effective communication may be neither appreciated nor overcome. Although expected benefits and risks may be briefly described, truly involving the patient in the decision-making process is often not a top priority (15). Patients frequently cannot recall crucial information about procedures within hours of giving consent (16-18). In two studies conducted more than a decade apart, more than 60% of patients surveyed about their experiences with the consent process said they believed that consent forms are intended to protect physicians' rights (19, 20).

So why did Ms. Morris sign the consent form? Could she have thought she was agreeing to the surgery to repair the second aneurysm that had been planned for a subsequent hospital stay? Perhaps. But if the electrophysiology fellow had thoroughly explained the electrophysiology study, it is difficult to imagine that the patient would have confused the two procedures. Ms. Morris did sign the form, but she clearly did not give “informed consent.”

9:00 to 9:15 a.m.Like the neurosurgery resident 90 minutes earlier, an interventional radiology attending went to Ms. Morris's room and was surprised to find it empty. He called the electrophysiology laboratory to ask why Ms. Morris was undergoing this procedure. The electrophysiology attending stated to the nurse that the call concerned a patient named Morris, but that Jane Morrison was on the table. The electrophysiology charge nurse corrected him, stating that, in fact, Joan Morris was on the table. The electrophysiology attending asked to see the patient's chart and recognized the error.

Comment from Ms. Morris upon the discovery of the error: And the next thing I know, I thought, “What is going on?” … when the doctor [the electrophysiology attending] says, “How do you spell your last name?” I thought, sometimes when you try to talk to somebody they don't listen, sometimes people won't listen to you. [Later] [T]he doctor who did the heart procedure came in. He said there would be a thorough investigation, and then he talked to my daughter … He apologized, and you could tell he was really shocked about it when he asked me my name and all.

To begin at the end, one of the many features of this case that deserves emphasis is the commendably immediate and complete disclosure of the error. We must overcome any temptation to be less than fully candid. The ethical imperative to inform patients and families when errors lead to adverse events overwhelms all other considerations.

On first reading, one may be tempted to blame this adverse event on any one of several individuals, from the nurse who mistakenly brought Joan Morris to the electrophysiology laboratory (RN₂) to the electrophysiology attending physician who failed to introduce himself to the patient at the start of the procedure. A closer analysis reveals problems beyond individual errors. To be sure, individuals made errors. In fact, discrete errors occurred in at least 17 different places. (See Appendix Table 1.) But this event shares many characteristics with other well-known and exhaustively researched calamities, such as the Challenger disaster, the Chernobyl nuclear reactor explosion, and the Bhopal chemical factory catastrophe. These events have been termed “organizational accidents” by psychologist and accident expert James Reason because they happen to complex, modern organizations, not to individuals (21, 22). No single individual error is sufficiently grave to cause an organizational accident. The errors of many individuals (“active errors”) converge and interact with system weaknesses (“latent conditions”), increasing the likelihood that individual errors will do harm.

Understanding why Ms. Morris mistakenly underwent the electrophysiology procedure requires looking beyond the actions of individuals to factors affecting the functioning of the systems in which the individuals acted. It is important to distinguish between two groups of these factors. Environmental factors are not readily changeable, at least in the short run, and thus they form the fixed context in which systems and people function. They act on all hospitals to increase the likelihood of this kind of adverse event. Latent conditions are system faults that can be remedied and act within individual hospitals to increase the probability that individuals will make errors, that errors will do harm, or both.

A disease analogy may clarify some of these relationships. Environmental factors are analogous to the genetic predispositions of an individual to develop atherosclerotic heart disease and its harmful sequelae. At present, these predispositions cannot be altered. Latent conditions resemble abnormalities such as hypertension or hypercholesterolemia. Like environmental factors, they can lurk unobserved for years, until the atherosclerotic plaque they promoted ruptures and causes a myocardial infarction. Unlike environmental factors, latent conditions can be effectively treated, reducing their capacity for harm.

Some of the most important environmental factors pertinent to this case are the increasing subspecialization in medicine (particularly in disciplines in which invasive procedures are an important part of practice), ongoing pressures to reduce hospital staffs, the trend to perform invasive procedures in hospitals on a short-stay basis, and the unremitting efforts of hospitals to reduce lengths of stay. These forces act on all hospitals to reduce the likelihood that an individual patient will be surrounded by physicians and nurses who know her well, understand why she is hospitalized, and actively coordinate planned tests and treatments. They act synergistically to increase the probability that the wrong patient will undergo an invasive procedure.

The most important latent conditions in this case include failures of communication, teamwork, and identity verification. Perhaps the most striking feature of this case—one that will be familiar to all clinicians who have worked in large hospitals—is the frighteningly poor communication it exemplifies. Physicians failed to communicate with nurses, attendings failed to communicate with residents and fellows, staff from one unit failed to communicate with those from others, and no one listened carefully to the patient. Although no data exist to document how widespread communication failures are, they are probably endemic in large, complex academic medical centers (23-30).

Poorly functioning teams are also a feature of this case. In addition to communicating well, effective teams allocate role responsibilities clearly, train to back up team members as necessary, monitor team members' performance, resolve conflicts efficiently, and use well-designed protocols and procedures to assure that complex tasks are executed flawlessly (31). Here, the oncology floor team failed in its responsibility to assure that Joan Morris received the care intended for her, and the electrophysiology laboratory team failed to keep track of whom they were treating and why.

How could so many well-trained and well-intentioned health care professionals ignore so many seemingly clear signals that they were subjecting the wrong patient to an invasive procedure? We recognize that retrospective root-cause analyses are susceptible to hindsight bias and may overestimate what it was reasonable for participants to know or anticipate in foresight (21). Nevertheless, we suspect that these physicians and nurses had become accustomed to poor communication and teamwork. A “culture of low expectations” developed, in which participants came to expect a norm of faulty and incomplete exchange of information. Nurses had probably observed many instances of patients' lacking information about planned procedures; RN₂ may have regarded Joan Morris's objections as just another such example. Similarly, residents may have grown accustomed to being unaware of all the tests or treatments ordered by attendings, and physicians may have often failed to fully inform nurses about treatment plans. The combined impact of these experiences probably led these conscientious professionals to discount the numerous warning signals present in this case. The culture of low expectations led each of them to conclude that these red flags signified not unusual, worrisome harbingers but rather mundane repetitions of the poor communication to which they had become inured.

What role did the similarity of the patients' names play in causing this adverse event? Patients with similar names present challenges to the best-functioning health care systems. In this case, the similar-sounding names led to errors that exposed long-standing system weaknesses that failed to prevent harm.

One of the most important defenses against this kind of adverse event was absent: a standardized protocol to verify patient identity (32). Despite the communication and teamwork failures, the adverse event could have been prevented at several different times if such a protocol had been in place and adhered to by either the electrophysiology laboratory or the oncology floor. Some automated verification systems (for example, bar-coding technology) may help to reduce the likelihood of misidentifications. But the technology still requires a protocol to be effective. A particular team member must be charged with matching the bar code on the patient's identity bracelet to the bar code on the medication, blood product, or invasive procedure schedule.

Furthermore, this hospital suffers from an information system disease that we suspect is common to many large academic medical centers: a patchwork of homegrown information minisystems, few of which interact effectively with each other. Because the electrophysiology laboratory's computer system could not connect to other hospital systems, it could not use their data to verify patient identities.

Finally, if Ms. Morris's medical record had contained legible and clear information about why she was in the hospital and what treatments were planned, one of her caregivers might have recognized the misidentification and averted this adverse event. The increasing frequency with which invasive procedures are performed during brief hospital stays encourages less documentation in the patient's medical record. Caregivers may thus expect little pertinent clinical information to appear in these patients' charts and not consider its absence worrisome.

Although these environmental factors and latent conditions were crucial in setting the stage for this event, individual factors undoubtedly also increased the likelihood of errors. We do not know all of the stressors that were operating on each of the individuals in this case, but a few common ones may have been involved. RN₂ was at the end of her shift on the oncology floor; she may have been in a hurry to leave and perhaps was less attentive to apparent warning signals than she would have otherwise been. Were other staff affected by this factor or by fatigue? Was the neurosurgery resident or electrophysiology fellow exhausted after a night on call (33-35)?

Factors that increase the likelihood of individuals making errors can never be completely eliminated. Human performance can be improved but not perfected. Industries that have reduced serious errors to extremely low levels have done so not by perfecting human performance but rather by improving the performance characteristics of the systems in which the humans work (22, 36, 37). Thus, the inevitable human errors are intercepted and prevented from doing harm. As Reason concludes, “We cannot change the human condition, but we can change the conditions under which people work” (21).

First, everyone practicing in complex delivery system settings should recognize that performing an invasive procedure on the wrong patient is an all-too-real possibility. No large hospital is immune from the individual errors or latent conditions present in this case. Yet, it appears that Joan Morris's caregivers did not conceive that it was possible that they had the wrong patient. As clinical teachers, we impress on trainees the importance of considering the most obscure diagnoses in evaluating individual patients. Similarly, we need to raise our index of suspicion for the possibility that patients are undergoing invasive procedures not intended for them.

Furthermore, we believe that open and vigorous discussion is a prerequisite for robust solutions. We were disappointed but not surprised that we could not find a single article in the literature discussing this problem. Given the environmental influences that are probably increasing the likelihood of these events, we must combat the clinical tunnel vision that subspecialization encourages. Nursing staffs should take particular care to familiarize themselves with short-stay patients and their treatment plans. No patient should leave a hospital floor for a procedure without a signed order and a fully executed consent form in the medical record. Hospitals should develop specific protocols for communicating vital clinical information when patients must spend time on inpatient units unfamiliar with their conditions. All units in which invasive procedures are performed must develop and adhere to routine, standardized procedures for verifying patient identity.

We believe that the communication and teamwork failures so prominent in this case are commonplace and lie at the root of many preventable adverse outcomes in health care delivery. Remedying such failures is at once our greatest challenge and our best hope for improvement. We should, however, resist the temptation to use punishment as an instrument of improvement in this case. No single error caused this adverse event; there is no reason to expect that punishing individuals would reduce the likelihood of recurrence.

Little research has addressed the relationship between communication or coordination of care and patient outcomes (38-40). We found no proven, effective interventions to improve communication and teamwork in health care delivery. However, a model for us to emulate does exist. When the National Aeronautics and Space Administration carefully studied the causes of airplane crashes in the 1970s, it concluded that 70% involved human error rather than irremediable mechanical failure. The most common errors related to failed communication and teamwork (41). These findings led to the development of a comprehensive set of training programs known as Crew Resource Management (CRM), which has now been implemented by the U.S. Federal Aviation Administration for all commercial airlines. These programs teach crews how to avoid barriers to effective communication and how to function well as teams. Evaluation has shown their effectiveness (42).

Recent research conducted by one of the developers of CRM has begun to characterize patterns of poor communication and teamwork among surgical and intensive care unit teams (31, 43). In one study, researchers compared the responses of pilots and surgical teams about several factors important in managing errors. Pilots were much more likely to acknowledge the adverse effects of fatigue on their own performance (64%) than were surgeons (18%) and to agree that junior team members should be free to question decisions of their seniors (97% vs. 55%) (44). Applying CRM to medicine will require the development, testing, and evaluation of methods to train (and periodically retrain) health care workers to value effective communication and teamwork, break down communication barriers (for example, hierarchies within and between professions, boundaries between departments), and function effectively as team members (for example, by repetitive practice of error management strategies in simulated patient care scenarios).

In this case, Joan Morris was mistakenly subjected to an invasive procedure over her repeated objections. Even though many individuals made errors, none was egregious or causative by itself. Instead, the systemic problems of poor communication, dysfunctional teams, and the absence of meticulously designed and implemented identity verification procedures permitted these errors to do harm. Just as we screen asymptomatic patients for hypertension, all health care systems should assess how well communication, teamwork, and protocols are functioning. Just as treating hypertension effectively prevents strokes, addressing underlying system flaws will greatly increase the likelihood that the inevitable errors of individuals will be intercepted and prevented from causing harm.

In the wake of Ms. Morris's experience, the Risk Management department convened an investigation and conducted a root-cause analysis. Several interventions directed at improving verification of patient identities, both within this particular department and throughout the hospital, were immediately instituted (Appendix Table 2).

Interviewer: We know that the mistake was discovered when the interventional radiology service went to round on their patient and found no patient and no chart. So let's move ahead to the issue that now confronts you as a risk manager in having to undertake the root-cause analysis for this mistake: How do you determine what you can fix? How will you fix it, or make recommendations as to what changes will be implemented, which will hopefully prevent this from happening again?

Ms. J: From my perspective, this is not the risk manager's job to fix. It is the risk manager's job to coordinate. It's very much a group process and the solution really has to come from the people involved in it because I can't tell them what they need to do. It doesn't work for them. They need to figure out how to fix their own area. For example, this e-mail schedule which they love because it is so fluid and it allows them to be flexible.

Interviewer: This is in the electrophysiology lab?

Ms. J: Yes. It's quite a problem because they don't always have first names, don't always have medical record numbers, they do tend to have clinical information, but the clinical information between Morris and Morrison was not that similar. So we have to convince them, actually they have to become convinced on their own, that what they need to do is move to a more formal system and insist on either using the medical record ID, or the date of birth, or to use the hospital's central computerized scheduling system, which is in place for that very reason. They believe [the centralized system] doesn't allow them the flexibility that they need. So we can't say to them, “You have to use the hospital's system. That's what risk management said and that's what you have to do.” You have to work with them, until they see they have to put a few more things into place.

No studies document the efficacy of measures to prevent patients from mistakenly undergoing invasive procedures. We base our discussion here on our assessment of the likelihood that specific interventions will remedy the system weaknesses that we believe are the underlying causes of the adverse event in this case. The improvements undertaken by the hospital are important first steps. Interventional Cardiology's new protocol to verify patient identity is a vital safeguard, as is the addition of birth dates or medical record numbers to patients' names in the scheduling system. Of even greater importance, the hospital has installed a strict protocol on all units to assure that only correctly identified patients with written orders in their charts are permitted to leave their floors for procedures. We encourage the hospital to require that a signed consent form also be present in the chart before a patient can be sent for an invasive procedure. Effective implementation of such a hospital-wide protocol will guard against wrong patient procedures in other parts of the hospital in addition to the electrophysiology laboratory.

Creating the protocols is a good start. Making them work effectively requires each affected unit to assign responsibilities clearly to specific caregivers for carrying out each step. We believe that at least two different people within Interventional Cardiology should independently use the protocol to verify patient identity. Nurses and physicians must be trained in the implementation of the protocols, and respected leaders must convince the staff that strict adherence must become routine. In addition, with experience the protocols should be revised to increase efficiency without sacrificing effectiveness, and periodic audits should be conducted to assess adherence.

We also believe the hospital should implement a plan to address the poor communication and teamwork evidenced by this case. Physician and nursing leaders should explicitly acknowledge that hierarchies within and between professions create undesirable barriers to effective exchange of information. They should develop a variety of multidisciplinary venues (for example, workshops, focus groups, patient rounds) to help overcome these obstacles. Discussions of this case in various parts of the hospital might provide the opportunity for physicians and nurses to generate their own ideas about how to improve communication and teamwork.

Finally, we believe that the hospital should increase the visibility of the informed consent process. All physicians who inform patients about procedures and obtain their consent should learn about this case. Ms. Morris' experience can serve as an effective reminder to physicians of their obligations to discuss procedures with patients thoroughly and in advance and to listen with care to their concerns.

Robert M. Wachter, MD (moderator): Dr. Chassin, what do you think the role of the malpractice system is in keeping these cases under the rug, and what should be changed to air them?

Dr. Chassin: Fear of malpractice litigation should not get in the way of this kind of discussion either externally—as we are doing today in a completely anonymous fashion—or, most importantly, internally, since virtually all quality improvement deliberations and actions are protected from discovery in every state. When a patient is unaware that an error has occurred, fear of a lawsuit may cause us to hesitate to disclose the error: “She doesn't know, what harm could result? It can only be made worse by disclosing it.” In my view, that reasoning is absolutely false. We clearly have an overriding ethical responsibility to disclose errors like this.

A Physician: Do patients have to become more proactive and vigilant?

Dr. Chassin: I am of two minds. First, I would advise prospective patients to become as well informed as possible—to know exactly which medications they are taking and which procedures are planned—in order to protect themselves. But my other reaction is: that's absurd. Why should we have to rely on patients to protect themselves? Hospitals ought to be the safest places in the world. We have a long way to go before we achieve that objective.