Disclosing Harmful Medical Errors to Patients: Tackling Three Tough Cases FREE TO VIEW

A growing number of articles in the literature^1–13 explores the issue of disclosing harmful medical errors to patients, emphasizing the importance of full disclosure and documenting the gap between this principle and current practice. One potential explanation for this gap is that health-care workers and institutions recognize that disclosure is “the right thing to do” but lack the moral courage to do what needs to be done.^14,15 An alternative explanation for this disclosure gap is that important challenges and complexities make it difficult for health-care workers to turn the general principle of disclosure into practice, obstacles that the existing literature^16–19 has not yet fully addressed.

The legal profession has long recognized that “hard cases make bad law”; in other words, deriving general principles from extraordinary cases can lead to rules that do not work well for more common dilemmas. Yet, analyzing difficult cases can reveal areas where the current thinking on disclosure may be underdeveloped and highlight the tradeoffs contained in seemingly straightforward disclosure cases.

In that spirit, we discuss three actual cases that present challenging disclosure dilemmas. The cases were collected from three health-care institutions in geographically diverse areas. Although details of the cases have been altered to maintain the anonymity of the patients, providers, and institutions, the essential aspects and resolutions of the cases have been preserved.

A middle-aged man was about to undergo a simultaneous kidney-pancreas transplant. As anesthesia was induced, the surgeon prepared the kidney and pancreas grafts for transplantation on the back table of the operating room by discarding fat, fascia, and accessory tissue. The surgeon implanted the kidney graft without difficulty, and the implanted kidney began producing urine. When the surgeon turned to retrieve the pancreas graft, he realized it was no longer on the back table. An extensive search for the pancreas graft ensued, including looking through the trash, but it could not be located. At this point, one of the operating room technicians realized that he had mistaken the pancreas graft for discarded debris on the back table and had disposed of it. He alerted the surgeon to what had happened. The operation was completed, and the patient returned to the recovery room in good condition.

After consultation with risk management, the surgeon decided to tell the patient's family that there was “good news and bad news” in that the kidney transplant had gone smoothly and the kidney was already producing urine but that the pancreas graft was “not transplantable,” so a second operation to implant a different pancreas graft would be required. If the patient's family inquired about why the pancreas graft was not transplantable, the surgeon planned to tell them that the pancreas was misidentified as debris and disposed of.

The surgeon approached the family in the waiting room and shared the information as planned. The family replied, “That's okay, doctor. What we really wanted was to get Dad off of dialysis; the pancreas transplant was just a bonus.” The family did not ask why the pancreas was not transplantable, and the surgeon did not offer additional details of what had transpired. The patient subsequently underwent a successful second operation to transplant a pancreas at the same institution.

This case represents a clear-cut, harmful error, one that existing guidelines say should be disclosed. Yet most guidelines^20–22 have been silent on what specific information should be conveyed. The standard of The Joint Commission²³ with regard to disclosure states that patients should be informed about all outcomes of care, including unanticipated outcomes, and frames the rationale for such disclosure as promoting informed decision making by the patient. However, the standard does not specify the content of the disclosure, leaving providers considerable discretion about how much information should be shared with the patient.

Uncertainty about exactly how much information to share with patients is one key barrier to disclosure.^16,17 Prior research^24,25 has suggested that patients value knowing specific information about the event, including whether an error occurred, why it occurred, and how it will be prevented in the future. Physicians, however, often begin disclosure conversations by providing very limited information about the event and relying on the patient or family to ask clarifying questions if they want to know more about what happened.¹⁷ This approach may partly reflect physicians' awareness that patients vary considerably in the amount of information they desire in other areas of clinical medicine. Physicians also might believe that this “don't ask-don't tell” strategy is consistent with a commitment to full disclosure. Indeed, the surgeon in this case was prepared to fully disclose everything that happened if the family asked why the pancreas was not transplantable. It is possible that the family did not ask follow-up questions because they considered why the pancreas was “not transplantable” to be unimportant and, thus, may have been satisfied with the disclosure as it occurred.

Yet, this strategy of letting the patient or family determine the content of disclosure through their questions has significant flaws. An alternative explanation for lack of follow-up questions could be that family members mistakenly believe that no error happened or that they are concerned that asking probing questions might cause problems for them (especially if they will need to rely on the same surgeon or institution for the second operation). Because health-care workers cannot read patients' or family members' minds, they cannot accurately intuit what the absence of questions means.

New disclosure standards increasingly describe the key information that should be shared, regardless of whether the patient or the family asks. For example, the guidelines for safe practices on disclosure of the National Quality Forum^10,26 require that patients be informed about the facts regarding the unanticipated outcome and its preventability. In the case at hand, sharing the fact that the failed pancreas transplant was preventable might have been all that was necessary for the family to want to learn more about the event.

Despite these emerging guidelines, clinicians may still wonder whether a specific piece of information needs to be disclosed to a patient. The medical profession should reach consensus about what constitutes material information in disclosure discussions (ie, information that should be shared with all patients regardless of whether the patient asks). One potential standard could be that specific disclosure content should be considered material if that information were essential for a reasonable patient or family to be free of fundamental misconceptions about what transpired.

Ensuring that patients are free of fundamental misconceptions about what happened not only is an important goal in and of itself, but also is essential for informed decision making broadly construed. Although the knowledge that an error occurred may not influence the patient's decision about whether to seek a second pancreas transplant, this information could help the patient to make an informed decision regarding whether to have the retransplantation at a different medical center. Our conception of materiality is consistent with the reasonable patient standard²⁷ that many state courts use retrospectively to assess the adequacy of informed consent. However, although legal standards for informed consent focus on the impact that information might have had on the patient's clinical decisions, disclosure of adverse events and errors can help patients to make more informed decisions about nonclinical issues as well. Knowledge of this error would likely help the patient and family to respond to bills received from the hospital, for example, reducing the chances that they would be financially harmed by this event.

The don't ask-don't tell approach in this case was deceptive. A reasonable family would not suspect that the pancreas was not transplantable because it had been discarded or even that an error occurred. Thus, the reason why the pancreas was not transplantable was material information that should have been shared, regardless of whether the patient or family asked.

An 83-year-old longstanding patient at a large medical center presented to the same center complaining of palpitations of 12 h duration. Neither her regular cardiologist nor her primary care physician was available, and she was seen by a covering cardiologist. An ECG showed atrial fibrillation. The finding was presumed to be new; therefore, she was scheduled for electrical cardioversion the following day without antecedent anticoagulation. The cardioversion was successful, and the patient was discharged home feeling well. The patient lived with her husband, who was in good health except for mild dementia. The couple had no children, and no other relatives lived locally.

Eighteen hours after arriving home, the patient experienced a massive embolic stroke and was taken to a different hospital, where she died. The patient's husband subsequently wrote a letter to his wife's regular physicians, informing them of her death and thanking them for providing her with many years of excellent health care. The letter prompted the primary care physician to review the old chart, as he recalled that the issue of atrial fibrillation might have arisen in the distant past. Indeed, there was an episode of atrial fibrillation documented 10 years previously. The patient was thought to be a fall risk, so anticoagulation was not prescribed at that time. Subsequent ECGs all had shown normal sinus rhythm, and the history of atrial fibrillation was not noted in her current problem list. The hospital debated at length whether to notify the patient's widower of this error in his wife's care, and ultimately decided that the distress such disclosure might cause the widower outweighed any potential benefit of disclosure.

This patient experienced the following serious medical error: failure to recognize that the atrial fibrillation was not new. Had the covering cardiologist discovered the history of atrial fibrillation, routine anticoagulation before cardioversion would have been recommended, and this would have likely prevented the subsequent fatal embolic stroke. Many arguments could be made to support the decision of the hospital not to disclose the error to the widower. Chief among them might be concern for his psychological well-being, underlying dementia, lack of surrogates, and the clinical uncertainty surrounding the fatal event (ie, the possibility that the embolic stroke might have been an independent event unrelated to cardioversion). In such situations, the reasons not to disclose can look very appealing. Cases in which an error is not apparent or in which health-care workers believe that patients or families might not understand the disclosure pose special problems for physicians, and data¹⁶ suggest that disclosure is less likely to happen in these instances. Other reasons may operate more subconsciously, for example, the desire to avoid a lawsuit, even if seemingly unlikely in this case, and the embarrassment of the error compounded by the irony of the widower's letter thanking the medical center for its excellent care.

This case invites careful consideration of the different rationales for disclosure and the way in which this reasoning can influence the disclosure decisions of institutions. As in the first case, one potential goal of disclosure is to promote informed decision making.²⁸ In this case, however, the hospital may have believed that because the patient died, no additional clinical decisions needed to be made. Although one could argue that disclosure could help the widower to make a more informed decision about whether to seek compensation, institutions typically view the possibility that disclosure could prompt litigation as a factor militating against, not in support of, disclosure. A second potential reason for disclosure is that discovery of error is inevitable, and being found to have “covered it up” places the physician and institution in a more compromised position than simply telling the truth. But, this rationale was not likely compelling in this case because it was considered unlikely that this error would be discovered.

The third potential rationale for disclosure centers on telling the truth for its own sake. But the value of telling the truth as an ideal when it is directly pitched against possible psychological harm without potential benefit poses a more difficult dilemma. The well-meaning decision makers at this institution were concerned by the possibility that the truth would do more harm than good, thus framing the lack of disclosure in benevolent terms. Benevolent deception, sometimes termed therapeutic privilege, historically has been considered an important part of medical practice. Some²⁹ have argued that “furthering a patient's ends may require withholding medical information,” and some^30,31 have suggested that benevolent deception in limited situations may be justifiable and distinct from lying.

Still, most see therapeutic privilege as a dangerous paternalism, that is, a slippery slope to be avoided, especially in today's medical environment that prioritizes patient autonomy and shared decision making. Data^32,33 from other areas, such as oncology and end-of-life care, suggest that patients want bad news even when it would be upsetting. Addressing the issue of therapeutic privilege, the American Medical Association³⁴ has taken a stance against withholding information, encouraging physicians to give patients the full story.

If the notion of benevolent deception has been generally discredited, why did it hold such sway in this disclosure decision? Ultimately, this case reflects two different types of deception. One, the widower was deceived about the care his wife had received, and two, the institution experienced self-deception regarding the relative benefits and burdens of disclosure. Such self-deception reveals an important flaw in the disclosure processes at many institutions, namely, the failure to recognize that disclosure of error reflects a conflict of interest that can lead institutions to subconsciously devalue or only partially consider the patient's interests. In the case at hand, the incomplete consideration of the patient perspective led the institution to conclude that the burdens of disclosure outweighed the benefits, whereas a more objective assessment would likely have concluded just the opposite.

A key consideration in the management of conflicts of interest is the awareness that conflicts of interest can distort judgment. Thus, we generally turn to a neutral party to assess whether a conflict exists and if so, how to proceed.²⁸ One option for obtaining such input would be to seek routine involvement of ethics committees in deliberations about disclosure. Close collaboration with the ethics committee could bring increased objectivity to the institution's disclosure decisions, especially with those ethics committees with lay members. The recommendations of ethics committees generally are advisory rather than binding, which could limit the influence of an ethics committee on the disclosure process.³⁵ However, given the sensitive nature of disclosure deliberations, the increased involvement of institutional ethics committees in this process could represent an important first step toward ensuring that the patient's perspective is more fully represented.

The medical profession should experiment with creative approaches to counteracting the understandable institutional impulse not to disclose in cases like this one, such as involving a patient or family advisory council, patient advocate, or other patient-representative voices. It would be neither realistic nor desirable to routinely hand off disclosure decisions to a third party or to exclude the involved health-care workers or institution. Yet, when the patient or family perspective gets lost, the decision-making process about disclosure suffers.

An additional institutional consideration is that disclosing the events to the husband would represent a commitment to transparency, sending a strong message to staff, and potentially the public, that transparency is valued at the institution, even when it is difficult or risks repercussions. Such a firm commitment to transparency is compelling but may create conflicts in cases in which consideration of the family member's psychological needs weigh against disclosure. In such cases, disclosure could potentially subordinate the family member's best interests to the pursuit of institutional benefits. Whether such disclosure honors the principles of a patient-centered approach and whether this factor is the most important metric for disclosure decisions altogether require more thoughtful attention as we shape future disclosure policies.

Institutions should be aware that self-deception can distort the judgments of even well-intentioned decision makers about disclosure and should ensure that the patient's voice is well represented in the disclosure process. Arguments that the nondisclosure of serious errors represents benevolent deception should be carefully scrutinized.

A 60-year-old patient underwent successful coronary artery bypass graft surgery. The attending surgeon left the operating room to speak with the family while the cardiac surgery fellow completed the skin sutures. The perfusionist handed the final bag of cell-saver blood to the anesthesiology resident for administration. Unbeknownst to the anesthesiology resident, the cell-saver bag contained a significant amount of air. The anesthesiology resident started to slowly infuse the blood. The patient's BP declined, and the attending anesthesiologist asked the resident to put the cell-saver blood under pressure to speed up its infusion. Ten minutes later, the patient had an asystolic cardiac arrest. The anesthesiology resident thought that he might have observed air in the IV line, which was quickly flushed with a normal saline solution.

The surgeon returned to find resuscitation underway. The anesthesiology resident revealed his concern that air may have been pushed into the IV line, information that guided the resuscitative effort. Fifteen minutes later, a normal BP was restored; however, the patient, previously awake and nearly ready for extubation, was no longer responsive to verbal or tactile stimuli.

The patient safety hotline and the patient communication consult service were notified, and advice was requested. The attending anesthesiologist and attending cardiac surgeon informed the family that there had been a cardiac arrest and that many explanations for the arrest were possible, including the accidental infusion of air through the IV line. They promised to conduct a full investigation and to remain in continuous contact with the family.

Overnight the intraoperative data were reviewed and confirmed that the likely presence of intracardiac air had caused the arrest. The following morning, the attending anesthesiologist informed the family of the known and suspected circumstances surrounding the injury, accepted responsibility for the error, and apologized. Within 24 h, the patient regained consciousness and was informed of the details surrounding the adverse event.

This case highlights how uncertainty may influence the disclosure process. For this injury, several aspects of uncertainty predominated, including whether an error had actually occurred, whether the error had harmed the patient (eg, whether the air embolism was etiologically responsible for the sudden deterioration), and how to characterize the patient's prognosis in the early stages following the arrest when he was not responsive to verbal or tactile stimuli, and long-term recovery therefore was unknown.

An intrinsic challenge to effective, timely disclosure is that full, accurate information about what happened may not be available immediately following the event. Physicians may feel uncomfortable about confronting the patient or family while armed with only partial information, a lack of an understanding of what happened, and uncertainty about his or her own responsibility for the event. Clinical uncertainty underscores the benefit of approaching disclosure as a process, not as an isolated event. The process can begin with an immediate disclosure that states what is known at the time, that an unanticipated outcome occurred, that it does or does not affect how the patient will be treated, and that an investigation is underway to determine why the injury occurred. The discussion can later continue with disclosures about the etiology of the injury and future prevention.

An initial real-time acknowledgment of an error or unanticipated outcome honors the obligation of telling the truth and enables the patient or family to make informed medical decisions. It also helps to avoid the distrust that may ensue from delayed or suppressed communication. The initial communication can include an assurance that additional information will be conveyed as it becomes available. At the same time, an opportunity for a full investigation relieves the involved providers of a belief that they have to have all the answers at the time of the initial family meeting. Connecting the patient and family with a liaison, as in this case, also demonstrates an ongoing commitment to open communication and support. A specific consult service with expertise in communication and crisis management can assist clinicians in preparing for disclosure discussions, provide a designated contact person for patients, and facilitate ongoing information sharing about the event and plans for follow-up.^36,37

Uncertainty about whether harm was caused by an error, however, can persist despite comprehensive investigation and analysis. No standards currently exist to guide clinicians' decisions in such cases. Should the provider share this uncertainty with the patient or family? How confident must the provider be that an error occurred in order for a disclosure obligation to be triggered?

From the patient's or family's perspective, ambiguity about whether the injury was attributable to an error will not seem a valid reason for not providing all the available information about what happened. The uncertainty does not change their need for information that may facilitate informed clinical decisions and alert them to the potential availability of an avenue for seeking compensation. When health-care providers include a candid statement of their level of certainty that an error occurred, they help the patient and family to weigh these matters. For example, hearing that the facts about injury causation are murky may dissuade the family from expending energy and resources on malpractice litigation that is unlikely to succeed.

The provider, on the other hand, may believe that admitting that an error may have occurred in the absence of persuasive evidence supporting causation simply opens a can of worms. Speculating about whether a mistake caused harm exposes the health-care team to litigation, damage to reputation, and potential severance of the physician-patient relationship without firm grounds. Although real, these concerns do not outweigh the patient's right to information. Physicians should work closely with institutional patient safety and quality experts to ensure that the information that they communicate to patients about a potential error is as accurate as possible without allowing such consultation to unduly delay disclosure discussions. Again, communicating the extent of uncertainty surrounding the injury and maintaining an open dialogue in an ongoing process of disclosure can help to minimize the risks to providers.

Uncertainty is inherent in the practice of medicine, and providers need help with understanding how to manage discussions about uncertainty with patients and families. Evolving disclosure policies must take this reality into account. Policies that demand a full disclosure in a matter of hours or days likely will result in an isolated disclosure conversation that is confined to what is reasonably knowable within that time frame, and often, that will exclude pieces of information that are important to patients and families, such as why an injury occurred and what will be done to prevent it from happening again. Approaching disclosure as a process, that is, coupling timely acknowledgment of the unanticipated outcome with full investigation and subsequent disclosure and apology as appropriate, is a preferable strategy.

The foregoing cases illustrate three of the central challenges facing health-care providers and architects of disclosure policies today. First, what information must be conveyed in a disclosure conversation? When is an omission misleading, and how much should be left up to the question-and-answer portion of a disclosure conversation rather than to the initial statement by the health-care provider? Second, how does the prospect that the disclosure itself might unproductively distress the patient or family affect a provider's disclosure obligations, and are providers and their institutions able to objectively weigh the prospect of patient distress given their own interest in avoiding admissions of error? Finally, what should be done when there is substantial uncertainty about whether an error occurred or whether it caused harm? How do the key rationales for disclosure (ie, the principle of telling the truth, the promotion of informed decision making, and the desirability of avoiding the potential harms associated with a cover-up) come into play in situations of uncertainty?

Although it may be undesirable to base policy decisions on hard cases, these three cases shed light on some of the hobgoblins of effective disclosure conversations. Disclosure decisions can be clouded by personal and institutional conflicts of interest and the very human temptation to answer unresolved questions (eg, Would this spouse want to know the truth? Did an error really occur?) in self-serving ways. By confronting these challenges at the time disclosure policies are created and disclosure training is conducted, institutions and providers can avoid ill-conceived and counterproductive decision making at the bedside later on and help providers to honor their professional obligations and their patients.

Financial/nonfinancial disclosures: The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article.

Other contributions: We thank Carolyn Prouty, DVM, and Odawni Palmer for assistance with manuscript preparation.