Error in Medicine: What Have We Learned?

MEDICAL WRITINGS Error in Medicine: What Have We Learned?

I don't watit to make the wrong mistake. in medicine has become more apparent, people Yogi Berra both inside and outside medicine have begun to examine medicine's systems and processes with this ver the past decade, it has become increasingly kind of "human factors" thinking in mind. Oapparent that error in medicine is neither rare In an early chapter of Human Error in Medicine, nor intractable. Traditionally, medicine has down- Leape reviews the data from the Harvard Medical played error as a negligible factor in complications Practice Study (5). This work was a landmark, in from medieal intervention. But, as data on the mag- large part because it was the methodologically strong- nitude of error aceumulate—-and as the public learns est study of its time to examine the epidemiology of more about them—medical leaders are taking the iatrogenic injury. The study assessed a random sam- issue seriously. In particular, the recent publication ple of patients discharged from New York hospitals of the Institute of Medicine report has resulted in in 1984. Serious iatrogenic injury occurred in 3.7 an enormous increase in attention from the public, percent of hospitalizations (6). On the basis of these the government, and medical leadership (1). results, it was estimated that iatrogenic injury con- tributed to as many as 180 000 deaths annually in the United States (7). This is four times the number Human Error in Medicine of deaths caused by traffic accidents. Other industries have done much better; although airplane Several books have been defining markers in this crashes get headlines, in 1998 no domestic airplane journey and highlight the issues that have emerged. fatalities occurred in the United States. Nonfatal Of particular note is Human Error in Medicine, ed- medical injuries resulting in disability or prolonged ited by Marilyn Sue Bogner (2), pubhshed in 1994 hospital stay occur in another 1.3 million U.S. pa- (unfortunately, currently out of print) and written tients per year (7). Even though medical profession- for those interested in error in medicine. Many of als conscientiously strive to avoid the "wrong mis- the thought leaders in the medical error field con- take," two thirds of medical injuries are estimated tributed chapters, and the contributions regarding to be preventable under the prevailing standard of human factors are especially strong. The book is a care (7). concise and clear introduction to the new paradigm In the human factors section of the book. Cook of systems thinking in medical error. and Woods discuss the difficulties that arise between In the foreword to this book. Reason provides in the "sharp end," practitioners who actually care for condensed form his thoughts about the evolution of patients, and those at the "dull end," persons who theory on the nature of human error and accidents control many of the resources (for example, admin- (3). There has been a transformation in thinking istrators) (8). They argue that although superficial and research about medical error and injury, much analyses are often focused at the sharp end, dull- of which has used information from other indus- end factors are more important than generally ap- tries. Conventionally, the single focus of both the preciated and contribute critically to most accidents. medical profession and the medieolegal system has Later chapters examine how such thinking has been on individual culpability for error. Other in- been applied in anesthesia (9, 10), which has been dustries, however^especially high-risk industries the leading specialty at addressing error systemati- such as aerospace and nuclear power—have pro- cally. Helmreich and Schaefer (9) discuss data on duced enormous improvements by focusing on re- the factors that influence team performance in the designing their systems to minimize human error. operating room and find that interpersonal and Reason, drawing from analysis of such disasters as communications issues play a key role, although the explosion of the space shuttle Challenger and most organizations devote little or no resources to the 1987 Kings Cross underground subway fire in support training in human factors and human inter- London, observes that systems often contain "latent ventions. The authors suggest that more attention to errors"—errors waiting to happen (4). Safeguards, the dull end could provide even better results; how- or "defenses," are often built in but are inadequate, ever, because of a variety of improvements, such as so that a series of failures can easily align to pro- oxygen monitoring, anesthesia has registered the duce disaster. As the extent of the problem of error largest safety gains of any specialty, Gaba—one of © 2000 American College of Physicians-American Society of Internal Medicine 763 the top researchers in this field—discusses the of well-designed medical systems. This comes across "chain of evolution of events" and how mistakes can clearly in Marianne Paget's The Unity of Mistakes: A often be recognized as such only in the context of a Phenomenological Interpretation of Medical Work specific, dynamic, complicated situation (10). (12), published in 1988, which probed the psyche of physicians on this topic. Paget analyzes in-depth interviews with 40 physicians and explores how phy- The Quality Perspective sicians think about their mistakes, as well as the psychological turmoil that physicians go through as The link between the patient safety movement they confront them. The book demonstrates how and the older, better-estabiishcd quality improve- difficult and common serious mistakes are for phy- ment movement has often been uncertain: At times, sicians, how hard each physician works to avoid the two groups have seemed to be entirely unaware mistakes, and yet (although Paget herself does not of one another. Journalist Michael Millenson's make this point) how limited individual effort is. highly acclaimed book, Demanding Medical Excel- Almost everyone in medicine works hard to avoid lence (11), published in 1997, shows how the two mistakes, and there does not seem to be much more disciplines fit together. Quality, he argues, should be room for individual effort to improve matters. An thought of as providing patients with the best pos- alternative path to progress is needed. sible medical care, and this requires continual progress not only at the highest levels of medical performance but also in shoring up the reliability Medicine's Changing Approach to Error and consistency of our most routine and basic tasks. While the book targeted the general public, most of The first Annenberg meeting in 1996—^whieh fo- those who have read it have some connection with cused on error in medicine and brought together a medicine. wide variety of disciplines—was a watershed event. The problems of error in medicine are addressed Many of the chapter authors in Bogner's book, in- in the chapter "First Do No Harm," and then pre- cluding Leape, Helmreich, and Cook and Woods, vention strategies are covered in "Saving Lives, Bit presented their view of the error "elephant." For by Byte." Millenson begins by providing a history of the first time, error experts from all over the world, the study of iatrogenic injury. He correctly points from both inside and outside medicine, convened in out that substantial data have been available for one place to discuss this issue. Many of the leading years. However, the publication of the results of the experts who participated were from different disci- Harvard Medical Practice Study finally provided plines and had not previously met. such detailed, incontrovertible evidence that the is- In subsequent years, as those studying error ex- sue could no longer be ignored. In the prevention amined risk and latent error in medical systems by chapter, he argues that information technology using perspectives that were new to many, the short- could allow the provision of much safer care by comings of the current approaches became obvious. linking disparate sources of medical knowledge and Most of the technologies and ways we do things in bringing them to clinicians at the time that chni- medicine were never designed with human limita- cians need them. In particular, he puts special em- tions in mind. Indeed, most medical processes were phasis on routinely incorporating outcomes data. At never consciously designed at all; rather, they were times, Millenson can be too strident, and he does built with a series of makeshift patches. Few such not get everything right—for example, he errone- processes in medicine would not benefit from a ously states that the Centers for Disease Control substantial overhaul. and Prevention has no definitions for iatrogenic infections— but this is perhaps the most readable, thorough discussion of the quality revolution in the Effective Prevention Strategies popular press, and it should be required reading for physicians. Early on, drug errors were identified as the most common type of medical error, and numerous research groups have since succeeded in redesigning Physicians' Reactions to Errors hospital pharmacy ordering and dispensing systems to markedly reduce in the incidence of drug errors. Much of the new paradigm is far off from the For example, implementation of unit dosing re- way physicians usually manage error. Traditionally, duced the frequency of medication errors by 82% in physicians have had a fierce ethic of personal re- one study (13). More recently, computerized physi- sponsibility, seeing conscientiousness as the key at- cian order entry—in which physicians enter pre- tribute of a good physician and good medicine—not scriptions online, where they can be checked for

764 2 May 2000 • Annals of Internal Medicine • Volume 132 • Number 9 problems—^was shown in a randomized trial to de- recently, the National Patient Safety Partnership crease the frequency of serious medication errors by was established, which includes both government 55% (14). In addition, having a clinical pharmacist organizations, such as the Department of Veterans follow an intensive care unit team on hospital Affairs, Department of Defense, and Health Care rounds was found to produce a 66% decrease in Financing Administration, as well as many private preventable adverse drug events (15). Researchers and professional organizations. The first activity of are now applying the tools of error analysis and this group was to emphasize the importance to human factors engineering more widely—to every- health care organizations of the "Y2K" problem, thing from surgical care to ambulatory medicine. In and the second has been to put forward a list of the area of fall prevention, for example, Tinetti and medication "best practices." colleagues (16) found that a multifactorial intervention to reduce the risk for falling among community-living elderly persons resulted in a decrease in risk for falling of almost one third (incident rate Discussion ratio, 0.69), and the intervention appeared to be cost-effective (17). Where We Are Now The sleeping giant has awoken. Both the public and purchasers are increasingly aware of the safety The Industry Response problems in medicine, and they are applying pres- sure. As a profession, we are at a crossroads. We The health care industry has begun to mobilize have solid epidemiologic data to demonstrate that to address these issues. The Joint Commission for iatrogenic injury is a major problem. Leaders are Accreditation of Hospital Organizations (JCAHO) now recognizing that the traditional response—that has altered investigations of major medical mishaps physicians do the best they can—is no longer enough. to focus not on blame and punishment of individu- And medicine has an increasingly robust array of als but on root cause analysis. For example., after a tools to improve the safety of care. Human factors Florida surgeon amputated a patient's wrong leg, and root cause analysis, in particular, offer ways to the JCAHO identified that wrong-site surgeries oc- dissect individual accidents and understand why curred at a steady rate across the United States; they occur (19). Methods that have been effective at they then developed process changes (marking the reducing error rates in other industries and appear operative site while the patient is awake) to avoid apphcable to medicine include simplifying, standard- the problem. The Institute for Healthcare Improve- izing, reducing unnecessary reliance on memory, im- ment has conducted several "breakthrough series" plementing forcing functions (reengineering a pro- for large groups of health care organizations on cess to prevent a specific error, such as requiring a medication errors and adverse drug events, which foot on the brake pedal to put a car in reverse), are intended to help hospitals learn effective strat- improving information access, reducing reliance on egies for preventing these events (18). These series vigilance, and reducing the number of handoffs in have resulted in important reductions in specific the system (18). Already, evidence suggests that these types of medication errors at many of the partici- and other strategies can dramatically reduce certain pating hospitals. types of error. In addition, major organizations are increasingly However, we are still only at the beginning. Al- focusing on this area. The American Medical Asso- though concrete measures for error reduction have ciation established the National Patient Safety been demonstrated, they have yet to be widely Foundation in 1997. A group that includes national adopted. Moreover, the information that does exist leaders in medicine as well as error experts has relates to only a few specific areas, such as adverse been convening at the John F. Kennedy School of drug events; even in this domain, the information Government at Harvard University for "Executive largely is only from inpatient care. By and large, Sessions on Medical Error and Patient Safety." The errors in medicine have gone unrecognized, unre- Massachusetts Medical Society has developed an ported, and unanalyzed. error reporting and feedback system, a superb but underfunded program that could serve as a model Where We Go From Here for enlightened regulation. Error reporting in this The first difficult step is making the transforma- system is confidential and nonpersonal, and there is tion to seeing medical systems and processes as the emphasis on getting hospitals to provide informa- critical source of most error, not individuals. That tion about what changes they made in follow-up; in requires moving away from a culture of fear about part because this medical society carries a "big admitting error and blame (20), and both the lead- stick," hospitals have been very responsive. And ership and individuals within the systems must make 2 May 2000 • Annals of Intemai Medicine • Volume 132 • Number 9 765 this journey. When other industries have success- Commission, for example, has suggested that the fully made the leap, they have not abandoned the Health Care Financing Administration consider pro- traditional ethic of personal responsibility but rather viding higher payment for institutions that imple- transformed it so that individuals take responsibility ment systems that have been demonstrated to im- for reporting error and fostering improvements. prove safety, such as physician order entry (26). In This takes strong leadership and conviction. Yet, addition, a group of the largest employers, dubbed unless errors and accidents are tracked on a routine the "Leapfrog Group," has banded together to drive basis and then used to improve care, we will be quality forward (27); this group is considering con- condemned to repeat the mistakes of the past. tracting preferentially or giving better rates to orga- Another pivotal problem is that complications nizations that adopt recommendations documented are persistently underestimated. No matter how to improve outcomes, with one of the first being protected individuals are, self-report is consistently computerized physician order entry in hospitals. unreliable. Chart review is effective but too expen- sive for routine use. There is some promise, however, that computerized event monitoring, in which Conclusion a program sits over a database and looks for signals that an adverse event may have occurred, will offer It is evident that the status quo is no longer an effective way of obtaining this information at a acceptable. Furthermore, we know much more reasonable cost (21-24). Such surveillance for no- about what to do about reducing the consequences socomial infections has been effective at reducing of error than we did, although it is still not nearly their incidence. Although to date such monitors enough. The next, painstaking task will be routinely have primarily been used to detect adverse drug measuring what we do, making the results publicly events, they will undoubtedly be used more broadly available, and then going through the systems of in the future—tracking, for example, pneumothorax medicine, one at a time, and bringing them forward rates after thoracentesis, as well as patterns of sur- to the 21st century. gical and bleeding complications. Finally, progress will require overcoming signifi- David W. Bates, MD, MSc cant organizational and financial obstacles. The task And A. Gawande, MD, MPH of reviewing existing medical practices using a sys- Brigham and Women's Hospital tems approach is daunting enough—few medical Boston, MA 02U5 processes would not benefit substantially from a "ground-up" overhaul. Success will require consid- Requests for Single Reprints: David W. Bates, MD, MSc, Division erable effort to change organization, staffing, train- of General Medicine and Primary Care, Brigham and Women's ing, and technology and, although safer systems Hospital, 75 Francis Street, Boston, MA 02115. tend to save resources over time, at least a mini- Requests To Purchase Bulk Reprints (minimum, 100 copies}: Bar- mum initial investment will be required (25). Un- bara Hudson, Reprints Coordinator; phone, 215-351-2657; e-mail, fortunately, during this turbulent time in medicine, [email protected]. severe financial pressures are forcing leaders to fo- Current Author Addresses: Dr. Bates: Division of General Medi- cus on short-term economic survival alone. As a cine and Primary Care. Brigham and Women's Hospital, 75 Francis Street, Bostim, MA 02115. result, safety is being left behind. Dr. Gawande: Department of Surgery, Brigham and Women's The traditional solution to this issue—^which has Hospital, 75 Francis Street, Boston, MA 02115. had serious problems—is regulation, by accrediting bodies such as the JCAHO. Although the JCAHO Ann Intern Med. 2000;!32:763-767. has substantially overhauled its approach to this issue and is now moving in a more optimal direc- tion, it has lagged behind quality efforts in other References areas, such as ISO 9000, a set of international qual- 1. Corrigan J, Kohn LT. Donaldson MS, eds. To Err is Human, Building a ity standards. Other regulatory bodies, such as the Safer Health 5yslem. Washington, DC: National Academy Pr; 2000. state boards for professionals, still often take an 2. Bogner MS, ed. Human Error in Medicine. Hillsdale, NJ: L. Eribaum Associ- ates, 1994. absurdly punitive tack, a notable example being the 3. Reason JT. Foreword. In; Bogner MS, ed. Human Error in Medicine. Hillsdale, NJ; L. Eribaum Associates; 1994:vii-xv. State Board of Nursing in Massachusetts, which 4. Reason Jr. Human Error. Cambridge, England; Cambridge Univ Pr; 1990. punished all 18 nurses who cared for a patient who 5. Leape LL. The preventabilily of medical injury. In: Bogner MS, ed. Human Error in Medicine. Hilbdale, NJ: L Eribaum Associates, 1994;13-25 suffered a chemotherapy overdose. 6. Brennan TA, Leape LL, Laird NM, Hebert L, Localio AR, Lawthers AG, An alternative to regulation for stimulating et al. Incidence of adverse events and negligence in hospitalized patients. Results from the Harvard Medical Practice Study I. N EngI J Med, 1991;324: health care organizations to improve the safety of 370-6. 7. Leape LL, Lawthers AG, Brennan TA, Johnson WG. Preventing medical the care they provide is for payers to provide finan- iniury. QRB Qual Rev Bull. 1993; 19; 144-9. cial incentives. The Medicare Payment Advisory 8. Cook R, Woods D. Operating at the sharp end: the complexity of human 766 2 May 2000 • Annals of Intemal Medicine • Volume 132 • Number 9 error. In: Bogner MS, ed. Human Error in Medicine. Hillsdale. NJ L. Eribaum 18. Leape LL, Kabcenell Al, Berwick DM. Reducing Adverse Drug Events, Associates; 1994:255-310. Boston: Institute for Healthcare Improvement; 1998, 9, Helmreich RL, Schaefer HQ. Team performance in the operating room. In: 19. Leape LL Error in medicine. JAMA. 1994;272:1851-7, Bogner MS, ed. Human Error in Medicine. Hillsdale, NJ: L, Eribaum Associates: 20. Kraman SS, Hamm G. Risk management: extreme honesty may be the best 1994:225-54. policy, Ann Intern Med. l999;131:963-7, 10. Gaba DM. Human error in dynamic medical domains. In: Bogner MS, ed. 21. Classen DC, Pestotnik SL, Evans RS, Burke JP. Computerized sun/eiilance Human Error in Medicine. Hillsdale, NJ: L, Eribsum Associates: 1994:197-224, of adverse drug events in hospital patients, JAMA. 1991:255:2847-51, 11. Millenson ML. Demanding Medical Excellence, Chicago: Univ of Chicago Pr; 22. Jha AK, Kupertnan GJ, Teich JM, Leape L, Shea B, Rittenberg E, et al. 1997, Identifying adverse drug events: development of a computer-based monitor 12. Paget MA. The Unity of Mistakes. Philadelphia: Temple Univ Pr; 1988. and comparison with chart review and stimulated voluntary report, J Am Med 13. Simborg DW, Derewicz HJ. A highly automated hospital medication sys- Inform Assoc. 1998;5:305-14, tem. Five years' experience and evaluation, Ann Intern Med. 1975:83:342-6, 23. Koch KE. Use of standardized screening procedures to identify adverse drug 14. Bates DW, Leape LL, Cullen DJ, Laird N, Petersen A, Teich JM, et al. reactions. Am J Hosp Pharm, 1990:47:1314-20, Effect of computerized physician order entry and a team intervention on 24. Honigman B, Bates DW, Light P. Computerized data mining for adverse prevention of serious medication errors. JAMA. 1998;280:1311-6, drug events in an outpatient setting [Abstraa], Proceedings of the American 15. Leape LL, Cullen DJ, Clapp MD, Burdick E. Demonaco HJ, Erickson Jl, Medical Informatics Fali Symposium; 7-11 November 1998, Orlando, Florida, et al. Pharmacist participation on physician rounds and adverse drug events Philadelphia: Hanley & Belfus; 1998.1018, in the intensive care unit, JAMA. 1999;282:257-70. 25. Teich JM, Glaser JP, Beckley RF, Aranow M, Bates DW, Kuperman GJ, 16. Tinetti ME, Baker Dl, McAvay G, Claus EB, Garrett P, Gottschalk M. et et al. Toward cost-effective, quality care; the Brigham Integrated Computing al. A multifactorial inten/ention to reduce the risk of falling among elderly System 2nd Nicholas E. Davies CPR Recognition Symposium, 1996:3, people living in the community. N EngI J Med. 1994;331:821-7. 26. The Medicare Payment Advisory Commission. Report to the Congress. 17. Rizzo JA, Baker Dl, McAvay G, Tinetti ME. The cost-effectiveness of a Selected Medicare Issues, Washington, D.C; 1999, multifactohal targeted prevention program for falls among community elderly 27. Bodenheimer T. The American health care system—the movement for im- persons, Med Care, 1995;34:954-69, proved quality in health care. N EngI J Med, 1999:340:488-92,

AD LIBITUM

Spring Fever

Timid in April, yellow-green buds Now wantonly coat the world With fairy powder, including my nose and eyes Which, ignoring the pharmacy on my windowsill, Insist it's May.

Harold E. Carlson, MD State University of New York at Stony Brook Stony Brook, NY 11794-8154

Requests for Single Reprints: Harold E. Carlson, MD, Endocrinol- ogy Divisioti, HSC T15-060, State University of New York, Stony Brook, NY 11794-8154,

2 May 2000 • Annals of Internal Medicine • Volume 132 * Number 9 767