A few weeks after the Annenberg Conference , Saul Weingart called me on the phone, introduced himself and said, “We should do an Executive Session on medical errors .” “What is an Executive Session?” I replied. He then told me about the work he had been involved in at the Harvard Kennedy School of Government (HKS) on juvenile justice and community policing. Developed in the late 1970s at HKS , an executive session is a prolonged confidential conversation among leaders in a practice field to solve a complex problem for which there is no evident technical solution.

In contrast to the usual approach to social problem-solving in which ideas are translated from research to practice, the executive session assumes that neither academics nor practitioners have the necessary knowledge, and therefore they must work together to create a solution. Sessions are ideally directed at problems that require changes in policy and management [1]. In a prior session on policing, for example, chiefs of police from major cities developed the concept of community policing.

Saul thought it was a promising model for the infant patient safety movement. I was intrigued from the start. Engaging the leaders of the major healthcare systems was key to getting patient safety moving. If we could get them and leaders of other national organizations to recognize the importance of patient safety, their own role, and the importance of developing meaningful interventions, we might be able to jump-start improvement for safety. This sounded like a good way to do it.

Saul had come to safety in a roundabout way. After graduating from Cornell, he got a PhD in public policy at the Kennedy School but then decided to go into medicine and graduated from the University of Rochester School of Medicine and Dentistry . During his internal medicine residency at Beth Israel Hospital, he got interested in quality improvement, which he could see as an organizational problem. He came across the announcement of the Annenberg Conference in JAMA and decided to apply. I think he was the only medical resident who attended the conference.

The Harvard politics were a bit sensitive, as the various schools (Medicine, Public Health, and Government) and affiliated teaching hospitals each had an interest and stake in the emerging field of patient safety. To navigate potential territorial conflicts, we sought out Joe Newhouse , the director of the Harvard-wide Division of Health Policy . A nationally known health economist, Joe agreed that his Division’s sponsorship would offer the proper auspices for the endeavor.

We formed a planning group that also included Miles Shore, a psychiatrist with long-standing interest in policy and leadership, Joe Newhouse , and the executive director of the Kennedy School Criminal Justice Center, Frank Hartmann , who was an architect of the executive session and a master facilitator.

We identified a diverse group of healthcare movers and shakers—individuals in positions of formal authority and important “influencers.” It turned out to be easier than I had expected to recruit key leaders. The people we were after were already aware that medical error was something that they had to deal with. With few exceptions, they readily signed on.

One of those exceptions was David Lawrence , head of Kaiser Permanente , who was initially cool to the idea. I thought that it was absolutely necessary to have him participate since Kaiser Permanente was both a premier provider of healthcare and highly organized, the type of healthcare organization that could more easily implement systems change . Because of its outstanding reputation, having it do so would send a powerful message to the rest of healthcare. So, although we had never met, I shamelessly “twisted his arm” over the phone, telling him he had to do it. Fortunately, he relented and came aboard. I later had the privilege of working with David on the IOM Quality of Care Committee , where his leadership was essential as we fashioned the legendary IOM reports.

figure 1

(a) Saul Weingart, (b) Miles Shore, (c) David Lawrence. (All rights reserved)

We were able to secure financial support from several sources: Agency for Health Research and Quality (AHRQ) , Veterans Health Administration , National Patient Safety Foundation , American Society of Health-System Pharmacists , and the WK Kellogg and Robert Wood Johnson Foundations .

Executive sessions follow a loose game plan. The early meetings are designed to draw out the participants by asking members to share personal stories of leadership success and failure. Discussions are topical, prompted by brief presentations and case studies, with the expectation that the discovery of lessons would be an iterative and collective process. Presenters and “reactors” are often drawn from the members. Some of the work of the session is done between meetings. And future agendas are built from the unresolved issues raised at earlier meetings. We would be making it up as we went along—which was the whole idea.

First Meeting, January 22–24, 1998

The first meeting of the Executive Session on Medical Error and Patient Safety convened January 22–24, 1998, at the Kennedy School. We had an unbelievable roster of participants , 30 in all—a “who’s who” of leadership in American healthcare delivery—exactly what we had hoped for. As well as leaders from key healthcare organizations, the group included business leaders, editors and news people, and academics. (See Appendix 7.1 for full list of members.)

After dinner, we welcomed everyone and explained what we were up to—the idea behind the executive session —and I gave the basic talk explaining the problem of medical error . The next day in the morning, we had two open discussions in which each member introduced himself and described examples of their leadership success and challenges. This helped create solidarity among the members and began a shared inventory of leadership lessons. Members also discussed the emerging recognition of patient safety as a silent epidemic. We agreed that CEOs are ultimately responsible for the persistence of error and are in the key position to do something about it.

In the afternoon, we had short talks, followed by respondents, followed by discussion. First, Mitch Rabkin , president and CEO of Beth Israel Deaconess Medical Center (BIDMC) , talked about barriers, following which members had an extensive discussion about the costs of error and fixing them. (There were little data on this at the time, but they were all keenly aware of it.) Next, Bob Frosch , former director of NASA, described his experience with NASA and other industries, with emphasis on principles of management, management by walking around, and the need for CEOs to think of healthcare as a production process. David Woods , systems engineering professor at Ohio State, spoke about error theory, the Swiss cheese model , and cognitive bias, especially hindsight bias.

Following dinner at the Harvard Faculty Club, John Nance , pilot and ABC News aviation correspondent, gave his exciting talk on aviation safety , emphasizing the value of crew resource management and communication against an authority gradient .

On Saturday, Don Berwick , president of the Institute for Healthcare Improvement (IHI) , summarized the previous day’s discussions and emphasized the need to adopt in healthcare lessons from human factors research. He then did something quite remarkable that proved to be fundamental to getting the CEOs really engaged: he gave the system CEOs homework. He challenged each of them to personally investigate an incident, including personally interviewing five or six involved participants , and to report back their experiences at the next meeting. He also asked them to invite a human factors specialist to spend a day in their institution and tell them their observations and, finally, to come back to the next meeting with a financial analysis of the cost of one error .

After an extensive discussion on the problems of confidential reporting because of the legal environment—something the group had interest in addressing—we concluded by members identifying a number of tasks for the group:

  • Define a working vocabulary for discussions about medical error.

  • Identify lessons from other industries.

  • Probe aviation for applicable lessons, such as confidential reporting and the value of a NASA-Ames research lab model for medical error.

  • Explore the relationship between error reduction and quality of care.

  • Identify areas of hazard or vulnerability in medicine.

  • Explore the impact of the tort system on reporting and the roles of oversight and regulatory organizations in maintaining accountability.

  • Identify constituencies whose voices should be represented at the exec. session.

  • Pilot test innovations at home.

The planning group was delighted with how it all went. We clearly had the interest and engagement of key leaders. Our job was to learn with them about how to move the needle on patient safety and to motivate them to take action. We thought we were off to a good start. We planned to have meetings every 6 months for 2–3 years.

Second Meeting: June 25–27, 1998

We had 24 in attendance, including 4 of the 5 who had missed the first meeting. We added a new member to the group: Timothy Johnson , medical editor, ABC News. Mark Moore of HKS gave the after-dinner talk on the role of the executive session and organizational strategy. The idea was to bring everyone, especially the new attendees, up to speed on what we are doing.

On Friday, Miles Shore gave an overview of the first meeting, and new members introduced themselves. I gave a presentation, Medical Error : Working Definitions. We then moved to the CEO’s reports on their experiences personally investigating a medical error in their own institutions. For all, it was eye-opening, sobering, and very motivating. Most of them had discovered things they had no idea were going on.

One member investigated an error in which a patient in the emergency department who was having a heart attack had a serious bleed after receiving a clot-busting thrombolytic drug. It turned out that the patient had received the wrong dose of thrombolytic, in part because it was difficult to obtain an accurate weight in the emergency department. On further investigation, it turned out that none of the last six patients with heart attacks had received the correct thrombolytic dose—and none of these cases had been previously identified or reported.

Another investigated a series of infections occurring among the patients of one plastic surgeon. After the second infection, nasal cultures were recommended for the whole operating room team, but the surgeon had refused to have one done . He was later required to have one and it was positive.

All of us learned how pervasive and serious the effects of errors are on the patients and on the staff. The research statistics we had presented earlier were jarring and interesting, but it was the personal stories that compelled these leaders toward resolve to change. Several CEOs were clearly moved by what they found—and that they had been so unaware of what was going on in their own institution. One described how he had reached out to a faculty member with expertise in human factors for advice. To his surprise, the faculty member said, “I’ve been waiting for your call for 20 years.”

Saul Weingart then gave a presentation describing high quality in the service sector, with examples from Taco Bell and the Ritz-Carlton. The goal was to translate promising approaches into concepts that might work in healthcare. We then had a discussion on putting worker safety first: a Harvard case study about Paul O’Neill , the CEO at Alcoa .

O’Neill had initiated a worker safety program at Alcoa in what was already the safest company in his industry. The case study raised for the group the linkage between keeping patients safe and keeping employees safe. Were there common principles? On Saturday, we discussed another fundamental concept in the construction of a safety culture : reporting. We considered confidential error-reporting systems , referring to the Billings work with the ASRS. Don then gave the challenge for the future.

But the big surprise of the meeting was provided by Jim Mongan , CEO of the Massachusetts General Hospital . He had missed the first meeting. After attending this one for 2 days, he announced that he was resigning because “I can’t tell my doctors how to practice.” We were stunned. Of course, what we were up to was pretty much what he said—not exactly telling doctors how to practice but trying to convince and inspire them to practice differently. That, of course, is what leadership is about. Interestingly, years later (post-IOM and after some good work by members of his own staff), Mongan became a strong advocate for safety. But he wasn’t there yet.

Following the meeting, at George Lundberg’s urging, six of us wrote an editorial that provided an update on patient safety activities, and he published in JAMA [2]. We discussed the application of human factors concepts to healthcare and described the work being done by NPSF , JCAHO , the Veterans Health Administration , and the Massachusetts Board of Registration in Medicine.

Third Meeting: January 21–23, 1999

Two new members were introduced: Jack Rowe , CEO of Mt. Sinai Hospital in New York, and Troy Brennan , representing Brigham and Women’s Hospital .

Our guest was Paul O’Neill , who gave the evening keynote, a compelling description of what he did to reduce harm at Alcoa . Bringing his case study to life, O’Neill emphasized that you cannot provide safe care if you don’t provide a safe workplace for your employees. O’Neill told the group that patient safety wasn’t a priority—it was a precondition. Paul became an outspoken advocate for patient safety and later joined the Lucian Leape Institute .

In the morning, members gave progress reports. There was no shortage of material! Bob Waller told us about the culture at Mayo, where the long-standing tradition is “the interest of the patient is the only interest.” Peter Van Etten described changes at UCSF to hold doctor’s accountable. Ken Kizer talked about the considerable changes he and Jim Bagian had made at the VA health system. These included medication bar coding , nonpunitive error reporting, removal of concentrated potassium chloride from floor stock, hazardous drug protocols, nursing upgrades, and assessing pain as the fifth vital sign.

Discussion followed on what needed to be done, and presentations were given on the lack of human factors testing for drug naming and labeling (a recognized contributor to medication error s), nonpunitive reporting , and building a culture of safety , followed by discussions focused on taking action.

Fourth Meeting: June 17–19, 1999

Don Berwick gave the evening keynote on Effective Executive Leadership . In the morning, Saul gave a recap of where we had come during the previous meetings of the executive session, and then we heard progress reports.

David Lawrence told about changes at K-P. He had launched major efforts at all levels—national, regional, and local. He writes a biweekly CEO journal on cases of harm and actions taken and shares best practices system-wide with all 70,000 members. He described his philosophy: no data without stories and no stories without data. But he also offered a cautionary note that effective messages require a relentless drumbeat and that people begin to understand what you mean at about the time when you are tired of saying it. He had stimulated a California statewide ADE initiative and was developing nationwide K-P accountability system. Joyce Clifford talked about nursing innovations at BIDMC .

Gordon Sprenger gave the most complete and inspiring account. He had engrafted patient safety into the governance of Allina , articulated safety as a major organizational goal, and integrated it into the strategic and operational plan. This included designating safety leaders, establishing a safety agenda, establishing the business case, and establishing nonpunitive reporting —including feedback about response. He personally inquired into medical accidents , coached senior managers in nonpunitive reporting , and kept talking with everyone about it.

He supported people who had made errors but had zero tolerance for violation of standards. He has changed the vocabulary from asking “who” to asking “what happened?” They were tapping 3-M engineers to help them with systems change. He disclosed plans to replicate the executive session for all CEO and Board chairs of Minnesota hospitals—which he subsequently did with the help of his safety leader, Julie Morath , and Saul Weingart .

figure 2

(a) Paul O’Neill and (b) Gordon Sprenger. (All rights reserved)

Sprenger, Lawrence, and Robert Waller had spoken out publicly on behalf of patient safety as well as fostering patient safety initiatives within their organizations. Jim Reinertsen , who had moved to become CEO of Boston’s CareGroup, made medication reliability one of four corporate priorities. Medication safety teams at each CareGroup hospital attempted to implement 16 best practices in 1 year and demonstrate measurable improvements in the safety of patients on anticoagulants and postoperative pain medications.

We then talked about how doctors and hospitals respond when a patient is harmed. Why are disclosure and apology so difficult? We showed the video from Annenberg of the Martin Memorial case, where the hospital stepped up and took responsibility for a death caused by a mistake. This was followed by a presentation by legal scholar Randy Bovbjerg on alternatives to the tort system and presentations by the executive director of National Alliance for the Mentally Ill, Laurie Flynn, on the consumer’s view; Sandy Fleming , executive director of MA Board of Registration in Medicine; and Dennis O’Leary , president of The Joint Commission on regulation and oversight. We had more discussion of actions that leaders could initiate. Overall, it was a very feisty meeting, with members fully engaged.

Fifth Meeting: January 27–29, 2000

This was the first meeting after the IOM report came out 2 months earlier, so we spent a whole session discussing it and its implications for leadership . We added two new members, Thomas Garthwaite , the new head of Veterans Health, replacing Ken Kizer , who had stepped down and Michael Wood , CEO of Mayo, replacing retiring Bob Waller . Jim Bagian gave the evening talk on his experience as an astronaut and now director of patient safety for the VA .

An interesting innovation was the presentation of a group of Harvard Business School-type case studies that Saul Weingart had commissioned with the goal of capturing the early experience of executive session members’ initiatives: creating a safety culture in the VA , creating a safety program at Allina , and patient safety at Mayo Clinic . Member reports included David Lawrence’s account of many new activities at Kaiser Permanente .

Lessons Learned

This was the last meeting of the executive session . We discussed how to spread the message and engage a broader swatch of leaders. We agreed on our key findings and principles, which in retrospect now seem remarkably applicable 20 years later. These were later summarized by Saul Weingart in testimony at the AHRQ Patient Safety Summit later in the year [3]:

  1. 1.

    Medical error is a problem of organizations. Members of the executive session embrace the view that medical error is an attribute of the systems and processes by which we deliver care. Scientific evidence and a wealth of experience from other industries demonstrate that human errors almost always result from defective systems. Improvement strategies that punish individual clinicians are misguided and do not work. Fixing dysfunctional systems, on the other hand, is the work that needs to be done.

  2. 2.

    Medical error is an executive responsibility. Because managers are responsible for organizing and shaping the systems and processes of care, hospital and health system executives share an essential and nondelegable responsibility for reducing medical error . Moreover, leadership lessons in patient safety can be learned and disseminated. To make progress, healthcare CEOs must commit their own time to working on behalf of patient safety. They must communicate its importance relentlessly. They must hold themselves personally accountable for patient safety in the same way they do for financial performance.

  3. 3.

    Many important lessons about high-reliability performance can be adapted from manufacturing, aviation, and the service sector. Executives should assess their organizations’ core processes for safety, inventory their organizations’ patient safety activities, report the results, implement best practices, and create a culture of safety .

  4. 4.

    Medical error is an urgent and strategic priority. Members of the executive session were impressed with the magnitude of harm represented by medical error . Ensuring that care is safe is a professional obligation for healthcare professionals and the organizations where they work. It is an urgent problem that requires the kind of immediate, focused, and sustained attention that motivated organizations to ensure Year 2000 (Y2K) compliance . Safety also makes good business sense. It builds consumer confidence and market share . Increased efficiencies and decreased rework may contribute to the bottom line.

  5. 5.

    Error reporting systems must be improved. Healthcare organizations must remain accountable to their patients and to the community by disclosing errors that result in harm , providing fair compensation for injuries, and introducing measures to prevent recurrence. Physicians have an ethical obligation to inform patients when they have been harmed because of an error in care.

  6. 6.

    Gross negligence and unethical behavior should not be shielded. Professional misconduct is a grave threat to patient safety and should be dealt with accordingly. But errors that do not result in harm must be protected from legal discovery, so that we can learn from them. Fear of discovery and punishment of clinicians’ accidents drives information underground and decreases organizational learning. We need to create robust nonpunitive error reporting systems. Sharing information about errors with frontline workers will build a sense of collaboration and shared mission.

  7. 7.

    The federal government should play an active role in patient safety, requiring pharmaceutical and device manufacturers to use human factors principles in naming, packaging, and labeling medications and to participate in post-market surveillance of adverse events .


Was the executive session a success? It did not create a national consensus or a comprehensive strategy for addressing patient safety. In retrospect, that would have been an unreasonable expectation, and it isn’t the purpose of executive sessions. Patient safety was in its infancy. Its definitions, methods, and scope were just developing. The major problems had yet to be defined. Patient safety was not ready for a grand strategy. But it was ready for big thinking.

Moreover, we were beginning to appreciate what an incredibly complex field patient safety is, bridging diverse disciplines from research to clinical care to administration, involving a broad range of stakeholders : patients, doctors, nurses, pharmacists, ancillary medical personnel, administrators, risk managers, lawyers, engineers, and government agencies. Achieving safe healthcare would require major changes in a physician-dominated culture that is more conservative and more authoritarian than in any institution in our society.

But the executive session was extraordinarily successful in other ways. While 99% of people working in patient safety have probably never even heard about it, the session educated major healthcare leaders in depth about patient safety and created awareness and understanding of its complexity . A community of concern developed in which leaders of major organizations in healthcare became colleagues in the pursuit of safe care. It motivated them to take action to advance the cause:

  • Dennis O’Leary led The Joint Commission to be more aggressive about safety, setting out National Patient Safety Goals and Standards.

  • David Lawrence made major changes at Kaiser Permanente that led it to become a model for patient safety among large healthcare systems.

  • Gordon Sprenger made Allina an exemplar of safety.

  • Steve Schroeder steered the Robert Wood Johnson Foundation to become the largest private funder of patient safety research and training.

An unanticipated effect of the executive session was its influence on observers. Although membership was by invitation, we encouraged participants to bring colleagues as observers. Two are of particular note. Gordon Sprenger brought Julie Morath , who went on to be a leader in patient safety as COO at Minnesota Children’s Hospitals and Clinics, and later as chief quality and safety officer at Vanderbilt University Medical Center and then as the president/CEO of the California Hospital Quality Institute, and member of the Lucian Leape Institute .

The other was Atul Gawande , a surgical resident who I had gotten to know during his year at the Harvard School of Public Health . Atul later developed the surgical checklist for WHO and created Ariadne Labs, an influential collaboration of innovators, implementers, and healthcare leaders focused on quality and safety. His books, Complications and Better, have succeeded more than any other in making safety issues accessible and understandable to the public.

The Harvard Executive Session concluded shortly after the IOM report was released and just as AHRQ and NQF began to play major roles in developing the foundations for the new field of patient safety (see chapters below). The executive session was complementary to these initiatives in that it helped develop the professional foundation: commitment by key leaders in the field. Patient safety is about changing systems of care. Systems leaders have to make that happen. The executive session set a stake in the ground early on, declaring that the most senior leaders in healthcare organizations had both the responsibility and capability to ensure safe care.

The young National Patient Safety Foundation recognized the potential of the executive session as a reproducible model for engaging leadership and motivating change. It subsequently sponsored executive sessions led by Saul Weingart in Minnesota and Indiana that brought together state hospital association leaders, health system CEOs , and (importantly) hospital trustees to navigate the emerging challenges of patient safety in their regions and communities [1]. Fierce economic competitors created a shared commitment to reducing medical errors and agreed to collaborate on patient safety interventions. This created a reservoir of good will and shared purpose among leaders in Minnesota and the regional Minnesota Alliance for Patient Safety that enabled bold interventions such as the 2003 first-in-the-nation Minnesota Adverse Health Events Reporting Law.