Professional Documents
Culture Documents
Lucian L. Leape, MD
FOR YEARS, medical and nursing stu- of omission or commission) or one that WHY IS THE ERROR RATE IN THE
dents have been taught Florence Night- does not achieve its intended outcome. PRACTICE OF MEDICINE SO HIGH?
ingale's dictum\p=m-\first,do no harm.1 Yet Indeed, injuries are but the "tip of the
evidence from a number of sources, re- iceberg" of the problem of errors, since Physicians, nurses, and pharmacists
ported over several decades, indicates most errors do not result in patient in¬ are trained to be careful and to function
that a substantial number of patients jury. For example, medication errors at a high level of proficiency. Indeed,
suffer treatment-caused injuries while occur in 2% to 14% of patients admitted they probably are among the most care¬
in the hospital.2-6 to hospitals,9"12 but most do not result in ful professionals in our society. It is cu¬
In 1964 Schimmel2 reported that 20% injury.13 rious, therefore, that high error rates
of patients admitted to a university hos- Aside from studies of medication er¬ have not stimulated more concern and
pital medical service suffered iatrogenic rors, the literature on medical error is efforts at error prevention. One reason
injury and that 20% of those injuries sparse, in part because most studies of may be a lack of awareness of the se¬
were serious or fatal. Steel et al3 found iatrogenesis have focused on injuries (eg, verity of the problem. Hospital-acquired
that 36% of patients admitted to a uni- the Harvard Medical Practice Study). injuries are not reported in the news¬
versity medical service in a teaching hos- When errors have been specifically papers like jumbo-jet crashes, for the
pital suffered an iatrogenic event, of looked for, however, the rates reported simple reason that they occur one at a
which 25% were serious or life threat- have been distressingly high. Autopsy time in 5000 different locations across
ening. More than half of the injuries studies have shown high rates (35% to the country. Although error rates are
were related to use of medication.3 In 40%) of missed diagnoses causing substantial, serious injuries due to er¬
1991 Bedell et al4 reported the results of death.1416 One study of errors in a medi¬ rors are not part of the everyday expe¬
an analysis of cardiac arrests at a teach- cal intensive care unit revealed an av¬ rience of physicians or nurses, but are
ing hospital. They found that 64% were erage of 1.7 errors per day per patient, perceived as isolated and unusual
preventable. Again, inappropriate use of which 29% had the potential for se¬ events—"outliers." Second, most errors
of drugs was the leading cause of the rious or fatal injury.17 Operational er¬ do no harm. Either they are intercepted
cardiac arrests. Also in 1991, the Har¬ rors (such as failure to treat promptly or or the patient's defenses prevent injury.
vard Medical Practice Study reported to get a follow-up culture) were found in (Few children die from a single misdi-
the results of a population-based study 52% of patients in a study of children agnosed or mistreated urinary infection,
of iatrogenic injury in patients hospi¬ with positive urine cultures.18 for example.)
talized in New York State in 1984.5·6 But the most important reason phy¬
Nearly 4% of patients suffered an injury For editorial comment see 1867. sicians and nurses have not developed
that prolonged their hospital stay or re¬ more effective methods of error pre¬
sulted in measurable disability. For New vention is that they have a great deal of
York State, this equaled 98 609 patients Given the complex nature of medical difficulty in dealing with human error
in 1984. Nearly 14% of these injuries practice and the multitude of interven¬ when it does occur.1921 The reasons are
were fatal. If these rates are typical of tions that each patient receives, a high to be found in the culture of medical
the United States, then 180000 people error rate is perhaps not surprising. The practice.
die each year partly as a result of iat¬ patients in the intensive care unit study, Physicians are socialized in medical
rogenic injury, the equivalent of three for example, were the recipients of an school and residency to strive for error-
jumbo-jet crashes every 2 days. average of 178 "activities" per day. The free practice.19 There is a powerful em¬
When the causes are investigated, it 1.7 errors per day thus indicate that hos¬ phasis on perfection, both in diagnosis
is found that most iatrogenic injuries pital personnel were functioning at a 99% and treatment. In everyday hospital
are due to errors and are, therefore, level of proficiency. However, a 1% fail¬ practice, the message is equally clear:
potentially preventable.4·7·8 For example, ure rate is substantially higher than is mistakes are unacceptable. Physicians
in the Harvard Medical Practice Study, tolerated in industry, particularly in haz¬ are expected to function without error,
69% of injuries were due to errors (the ardous fields such as aviation and nuclear an expectation that physicians translate
balance was unavoidable).8 Error may power. As W. E. Deming points out (writ¬ into the need to be infallible. One result
be defined as an unintended act (either ten communication, November 1987), is that physicians, not unlike test pilots,
even 99.9% may not be good enough: "If come to view an error as a failure of
we had to live with 99.9%, we would have: character—you weren't careful enough,
From the Department of Health Policy and Manage- 2 unsafe plane landings per day at O'Hare, you didn't try hard enough. This kind of
ment, Harvard School of Public Health, Boston, Mass.
Reprint requests to Department of Health Policy and
16000 pieces of lost mail every hour, thinking lies behind a common reaction
Management, Harvard School of Public Health, 677
32000 bank checks deducted from the by physicians: "How can there be an
Huntington Ave, Boston, MA 02115 (Dr Leape). wrong bank account every hour." error without negligence?"
their own challenges to system design: conferences, incident reports, risk man¬
exercise high level cognitive skills in a most
complex domain about which much is known, sensory overload and boredom. None¬ agement activities, and quality assur¬
theless, these safeguards have served ance committees abound. But, as noted
but where much remains to be discovered.31
the cause of aviation safety well. previously, these activities focus on in¬
While the comparison is apt, there Second, procedures are standardized cidents and individuals. When errors are
are also important differences between to the maximum extent possible. Specific examined, a problem-solving approach
aviation and medicine, not the least of protocols must be followed for trip plan¬ is usually used: the cause of the error is
which is a substantial measure of un¬ ning, operations, and maintenance. Pilots identified and corrected. Root causes,
certainty due to the number and variety go through a checklist before each take¬ the underlying systems failures, are
of disease states, as well as the unpre¬ off. Required maintenance is specified in rarely sought. System designers do not
dictability of the human organism. detail and must be performed on a regu¬ assume that errors and failures are in¬
Nonetheless, there is much physicians lar (by flight hours) basis. Third, the evitable and design systems to prevent
and nurses could learn from aviation. training, examination, and certification or absorb them. There are, of course,
Aviation—airline travel, at least—is process is highly developed and rigidly, exceptions. Implementation of unit dos¬
indeed generally safe: more than 10 as well as frequently, enforced. Airline ing, for example, markedly reduced
million takeoffs and landings each year pilots take proficiency examinations ev¬ medication dosing errors by eliminating
with an average of fewer than four ery 6 months. Much of the content of ex¬ the need for the nurse to measure out
crashes a year. But, it was not always aminations is directly concerned with each dose. Monitoring in intensive care
so. The first powered flight was in 1903, procedures to enhance safety. units is sophisticated and extensive (al¬
the first fatality in 1908, and the first Pilots function well within this rigor¬ though perhaps not sufficiently redun¬
midair collision in 1910. By 1910, there ously controlled system, although not dant). Nonetheless, the basic health care
were 2000 pilots in the world and 32 had flawlessly. For example, one study of system approach is to rely on individu¬
already died.32 The US Air Mail Service cockpit crews observed that human er¬ als not to make errors rather than to
was founded in 1918. As a result of ef¬ rors or instrument malfunctions oc¬ assume they will.
forts to meet delivery schedules in all curred on the average of one every 4 Second, standardization and task de¬
kinds of weather, 31 of the first 40 Air minutes during an overseas flight.32 sign vary widely. In the operating room,
Mail Service pilots were killed. This ap¬ Each event was promptly recognized it has been refined to a high art. In
palling toll led to unionization of the pi¬ and corrected with no untoward effects. patient care units, much more could be
lots and their insistence that local field Pilots also willingly submit to an exter¬ done, particularly to minimize reliance
controllers could not order pilots to fly nal authority, the air traffic controller, on short-term memory, one of the the
against their judgment unless the field when within the constrained air and weakest aspects of cognition. On-time
controllers went up for a flight around ground space at a busy airport. and correct delivery of medications, for
INSTITUTIONALIZATION OF SAFETY
basic tenants of total quality manage¬ cept the notion that error is an inevi¬
ment, statistical quality control, requires table accompaniment of the human con¬
Although the idea of a national hos¬ data regarding variation in processes. dition, even among conscientious pro¬
pital safety board that would investi¬ In a generic sense, errors are but varia¬ fessionals with high standards. Errors
gate every accident is neither practical tions in processes. Total quality man¬ must be accepted as evidence of sys¬
nor necessary, at the hospital level such agement also requires a culture in which tems flaws not character flaws. Until
activities should occur. Existing hospi¬ errors and deviations are regarded not and unless that happens, it is unlikely
tal risk management activities could be as human failures, but as opportunities that any substantial progress will be
broadened to include all potentially in¬ to improve the system, "gems," as they made in reducing medical errors.
jurious errors and deepened to seek out are sometimes called. Finally, total qual-
References
1. Nightingale F. Notes on Hospitals. London, Eng- 79:718-722. 25. Deming WE. Quality, Productivity, and Com-
land: Longman, Green, Longman, Roberts, and 13. Bates DW, Boyle D, Vander Vliet M, et al. petitive Position. Cambridge: Massachusetts Insti-
Green; 1863. Relationship between medication errors and ad- tute of Technology; 1982.
2. Schimmel EM. The hazards of hospitalization. verse drug events. J Gen Intern Med. In press. 26. Reason J. Human Error. Cambridge, Mass:
Ann Intern Med. 1964;60:100-110. 14. Anderson RE, Hill RB, Key CR. The sensitiv- Cambridge University Press; 1992.
3. Steel K, Gertman PM, Crescenzi C, et al. Iatro- ity and specificity of clinical diagnostics during five 27. Rasmussen J, Jensen A. Mental procedures in
genic illness on a general medical service at a uni- decades: toward an understanding of necessary fal- real-life tasks: a case study of electronic trouble\x=req-\
versity hospital. N Engl J Med. 1981;304:638-642. libility. JAMA. 1989;261:1610-1611. shooting. Ergonomics. 1974;17:293-307.
4. Bedell SE, Deitz DC, Leeman D, Delbanco TL. 15. Goldman L, Sayson R, Robbins S, Conn LH, 28. Norman DA. To Err Is Human. New York,
Incidence and characteristics of preventable iatro- Bettman M, Weissberg M. The value of the autopsy NY: Basic Books Inc Publishers; 1984.
genic cardiac arrests. JAMA. 1991;265:2815-2820. in the three medical eras. N Engl J Med. 1983;308: 29. Tversky A, Kahneman D. The framing of de-
5. Brennan TA, Leape LL, Laird N, et al. Inci- 1000-1005. cisions and the psychology of choice. Science. 1981;
dence of adverse events and negligence in hospi- 16. Cameron HM, McGoogan E. A prospective study 211:453-458.
talized patients: results of the Harvard Medical of 1,152 hospital autopsies, I: inaccuracies in death 30. Yerkes RM, DodsonJD. The relation of strength
Practice Study I. N Engl J Med. 1991;324:370-376. certification. J Pathol. 1981;133:273-283. of stimuli to rapidity of habit formation. J Comp
6. Leape LL, Brennan TA, Laird N, et al. The 17. Gopher D, Olin M, Donchin Y, et al. The nature Neurol Psychol. 1908;18:459-482.
nature of adverse events in hospitalized patients: and causes ofhuman errors in a medical intensive care 31. Allnutt MF. Human factors in accidents. Br J
results of the Harvard Medical Practice Study II. unit. Presented at the 33rd annual meeting ofthe Hu- Anaesth. 1987;59:856-864.
N Engl J Med. 1991;324:377-384. man Factors Society; October 18,1989; Denver, Colo. 32. Perrow C. Normal Accidents: Living With High\x=req-\
7. Dubois RW, Brook RH. Preventable deaths: who, 18. Palmer RH, Strain R, Rothrock JK, et al. Evalu- Risk Technologies. New York, NY: Basic Books
how often, and why? Ann Intern Med. 1988;109: ation of operational failures in clinical decision mak- Inc Publishers; 1984.
582-589. ing. Med Decis Making. 1983;3:299-310. 33. Orkin FK. Patient monitoring during anesthe-
8. Leape LL, Lawthers AG, Brennan TA, Johnson 19. Hilfiker D. Facing our mistakes. N Engl JMed. sia as an exercise in technology assessment. In:
WG. Preventing medical injury. Qual Rev Bull. 1984;310:118-122. Saidman LJ, Smith NT, eds. Monitoring in Anes-
1993;8:144-149. 20. Christensen JF, Levinson W, Dunn PM. The thesia. 3rd ed. London, England: Butterworth Pub-
9. Lesar TS, Briceland LL, Delcoure K, et al. Medi- heart of darkness: the impact of perceived mistakes lishers Inc; 1993.
cation prescribing errors in a teaching hospital. on physicians. J Gen Intern Med. 1992;7:424-431. 34. Gaba DM. Human errors in anesthetic mishaps.
JAMA. 1990;263:2329-2334. 21. Wu AW, Folkman S, McPhee SJ, et al. Do house Int Anesthesiol Clin. 1989;27:137-147.
10. Raju TN, Thornton JP, Kecskes S, et al. Medi- officers learn from their mistakes? JAMA. 1991; 35. Cooper JB, Newbower RS, Kitz RJ. An analy-
cation errors in neonatal and paediatric intensive\x=req-\ 265:2089-2094. sis of major errors and equipment failures in an-
care units. Lancet. 1989;2:374-379. 22. Berwick DM. E. A. Codman and the rhetoric of esthesia management: considerations for preven-
11. Classen DC, Pestonik SL, Evans RS, Burke JP. battle: a commentary. Milbank Q. 1989;67:262-267. tion and detection. Anesthesiology. 1984;60:34-42.
Computerized surveillance of adverse drug events 23. McIntyre N, Popper KB. The critical attitude 36. Cullen DJ, Nemeskal RA, Cooper JB, Zaslavsky
in hospital patients. JAMA. 1991;266:2847-2851. in medicine: the need for a new ethics. BMJ. 1989; A, Dwyer MJ. Effect of pulse oximetry, age, and
12. Folli HL, Poole RL, Benitz WE, Russo JC. 287:1919-1923. ASA physical status on the frequency of patients
Medication error prevention by clinical pharma- 24. Berwick DM. Continuous improvement as an admitted unexpectedly to a postoperative care unit.
cists in two childrens' hospitals. Pediatrics. 1987; ideal in health care. N Engl J Med. 1989;320:53-56. Anesth Analg. 1992;74:181.