Pierluissi E, Fischer MA, Campbell AR, Landefeld CS. Discussion of Medical Errors in Morbidity and Mortality Conferences. JAMA. 2003;290(21):2838-2842. doi:10.1001/jama.290.21.2838
Author Affiliations: San Francisco Veterans Affairs Medical Center (Drs Pierluissi and Landefeld), San Francisco General Hospital (Dr Campbell), and Departments of Surgery (Dr Campbell) and Epidemiology and Biostatistics (Dr Landefeld), School of Medicine, University of California, San Francisco; and Palo Alto Veterans Affairs Medical Center, Stanford University, Palo Alto, Calif (Dr Fischer). Dr Pierluissi is now with San Mateo Medical Center, San Mateo, Calif. Dr Fischer is now with the Division of Primary Care, University of Massachusetts, Worcester.
Context Morbidity and mortality conferences in residency programs are intended
to discuss adverse events and errors with a goal to improve patient care.
Little is known about whether residency training programs are accomplishing
Objective To determine the frequency at which morbidity and mortality conference
case presentations include adverse events and errors and whether the errors
are discussed and attributed to a particular cause.
Design, Setting, and Participants Prospective survey conducted by trained physician observers from July
2000 through April 2001 on 332 morbidity and mortality conference case presentations
and discussions in internal medicine (n = 100) and surgery (n = 232) at 4
US academic hospitals.
Main Outcome Measures Frequencies of presentation of adverse events and errors, discussion
of errors, and attribution of errors.
Results In internal medicine morbidity and mortality conferences, case presentations
and discussions were 3 times longer than in surgery conferences (34.1 minutes
vs 11.7 minutes; P = .001), more time was spent listening
to invited speakers (43.1% vs 0%; P<.001), and
less time was spent in audience discussion (15.2% vs 36.6%; P<.001). Fewer internal medicine case presentations included adverse
events (37 [37%] vs 166 surgery case presentations [72%]; P<.001) or errors causing an adverse event (18 [18%] vs 98 [42%],
respectively; P = .001). When an error caused an
adverse event, the error was discussed as an error less often in internal
medicine (10 errors [48%] vs 85 errors in surgery [77%]; P = .02). Errors were attributed to a particular cause less often in
medicine than in surgery conferences (8 [38%] of 21 medicine errors vs 88
[79%] of 112 surgery errors; P<.001). In discussions
of cases with errors, conference leaders in both internal medicine and surgery
infrequently used explicit language to signal that an error was being discussed
and infrequently acknowledged having made an error.
Conclusions Our findings call into question whether adverse events and errors are
routinely discussed in internal medicine training programs. Although adverse
events and errors were discussed frequently in surgery cases, teachers in
both surgery and internal medicine missed opportunities to model recognition
of error and to use explicit language in error discussion by acknowledging
their personal experiences with error.
Recognition of the inevitability of errors in fields such as aviation,
nuclear power generation, and electronics manufacturing has been followed
by impressive quality improvement and error reduction.1 In
medicine, incorporation of this knowledge has been slow,2 despite
the increased attention of professional,3 governmental,4 and private5 organizations
following the 1999 Institute of Medicine4 report
exposing the magnitude of the medical error problem. These groups believe
that increased reporting and study of errors will lead to system improvements
and safer health care. Openness to discussion and study of errors, with a
realization that "errors must be accepted as evidence of systems flaws, not
character flaws"1 is central to their message.
The morbidity and mortality conference is one of academic medicine's
most visible forums for discussion of adverse events and errors.6- 8 It
is unknown, however, to what extent adverse events and errors are discussed
in morbidity and mortality conferences.
We sought to answer the question of whether errors are discussed in
medicine and surgery morbidity and mortality conferences. We studied this
conference at 4 teaching hospitals affiliated with 2 US medical schools. Our
research questions were the following: (1) Are adverse events and errors presented
in medicine and surgery morbidity and mortality conferences? (2) Are adverse
events and errors discussed in these conferences? and (3) How are errors discussed
in these conferences?
We conducted a cross-sectional study of morbidity and mortality conferences
in the departments of medicine and surgery at San Francisco General Hospital,
San Francisco Veterans Affairs Medical Center, University of California at
San Francisco Medical Center, and Stanford University Medical Center. Conferences
were held weekly or monthly and were attended by medical students, residents,
and faculty. We identified 247 consecutive conferences from July 2000 until
April 2001; based on availability of a study team member, 151 conferences
(61%) were attended, including 66 medicine conferences in which 100 individual
cases were presented, and 85 surgery conferences in which 232 individual cases
Measures were developed based on the taxonomy and methods of the Harvard
Medical Practice Study.9 Using structured implicit
review, each case presentation and discussion was assessed by a trained physician
observer to determine the occurrence of an adverse event, the occurrence of
an error, whether the error caused the adverse event, and the scope of discussion.
An adverse event was defined as an unintentional,
definable injury that was the result of medical management and not a disease
process. An error was defined as the failure of a
planned action to be completed as intended or the use of a wrong plan to achieve
an aim.10 The physician observers used a 6-point
ordinal scale9 to rate the evidence that an
adverse event occurred, that an error occurred, and that the error caused
the adverse event. For our analysis, we determined that an error caused an
adverse event when the observer's judgments of the probability that an adverse
event occurred, that an error occurred, and that the error caused the adverse
event were each greater than 50%.
A discussion was defined as a meaningful exchange between 2 or more
speakers involving a comment, a response, and a follow-up response. A single
statement about an error was not considered a discussion. The discussion of
error in the conference was classified in 1 of 6 categories: no discussion,
explicit discussion (using the words "error," "mistake," or equivalent), implicit
discussion (without use of the words "error," "mistake," or equivalent), a
disagreement between physicians, a nonspecific problem, or a reasonable choice.
When an error was discussed, it was determined whether the participants attributed
the error to an individual, a team, a system, or a combination of these.
Physician observers underwent extensive structured training, attended
the conferences, applied the measures as described, and collected demographic
data on each case including age, sex, and mortality outcome. They also measured
time allotted to case presentation, conference leader speaking, invited lecturer
speaking, and audience participation.
Medicine conferences were observed by 1 of 4 board-eligible internists,
and surgery conferences were observed by 1 of 4 fourth-year surgery residents
or a third-year surgery resident. Observer assessments were made after the
case presentation and discussion were complete. Because we wished to study
what transpired in the conferences alone, observers did not review medical
records, participate in the conferences, or discuss cases.
We determined reliability of assessments by measuring interrater agreement
among multiple observers at the same conference using the κ statistic.
For medicine, 2 to 3 observers simultaneously attended 10 medicine cases;
for surgery, 2 to 4 observers simultaneously attended 24 surgery cases. For
medicine cases, κ values and agreement were 0.85 (95% confidence interval
[CI], 0.35-1.00) and 96% for the presence of an adverse event and 0.78 (95%
CI, 0.28-1.00) and 96% for the presence of an error. For surgery cases, κ
values and agreement were 0.58 (95% CI, 0.38-0.77) and 89% for the presence
of an adverse event and 0.70 (95% CI, 0.50-0.89) and 93% for the presence
of an error.
Comparisons between departments for categorical variables were made
using the χ2 statistic, with the Fisher exact test for counts
of less than 5. Comparisons with continuous variables were made using the t test. Proportions were compared using the z test. P<.05 was considered significant.
Analyses were performed with Stata software, version 6.0 (Stata Corp, College
This study was performed with approval of the human subjects committees
at the participating institutions. Permission to observe and rate the morbidity
and mortality conferences was obtained from each department's chair. Conference
leaders and participants were aware that the conference was being evaluated
but were unaware of the specific evaluation.
Cases presented in medicine and surgery morbidity and mortality conferences,
respectively, were similar in mean age (52 and 48 years; P = .13), proportion of women (24% and 28%; P =
.47), and proportion who died (23% and 24%; P = .82).
On average, fewer cases were presented in medicine conferences than in surgery
conferences (1.5 cases vs 2.7 cases; P = .001) and
the average time spent on each case was 3 times longer for medicine cases
(34.1 minutes vs 11.7 minutes; P = .001). In medicine
conferences, most time was allotted to presenting the case and listening to
invited speakers, with relatively little audience participation (Table 1). In contrast, surgery conferences
focused on case presentation and audience discussion.
Adverse events were present in 37 case presentations (37%) in medicine
conferences compared with 166 presentations (72%) in surgery conferences (P<.001) (Figure 1).
Errors resulting in an adverse event were present in 18 medicine cases (18%)
compared with 98 surgery cases (42%; P = .001) (Figure 1). There was no difference in the
proportion of adverse events associated with an error in medicine and surgery
conferences (48% and 59%, respectively; P = .24).
Deaths due to an error occurred in 5 medicine cases and 14 surgery cases.
Whereas only 11 (52%) of 21 errors presented in medicine conferences occurred
on the medicine service, nearly all errors (106/113 [95%]) presented in surgery
conferences occurred on the surgery service (P<.001).
Error discussion occurred in 10 (10%) of 100 medicine cases and in 80
(34%) of 232 surgery cases (P<.001 for difference
between medicine and surgery). In medicine conferences, when errors were presented,
they were less likely to be discussed as an error and more likely to be ignored
altogether compared with surgery conferences (Table 2). In both medicine and surgery conferences, when errors
were discussed as errors, only 40% were discussed explicitly. Of the 7 errors
associated with 5 deaths in medicine, 3 (43%) were not discussed. Of the 17
errors associated with 14 deaths in surgery, only 1 (6%) was not discussed
(P = .06). See Table 3 for examples of cases with fatal and nonfatal errors.
Of the 10 cases with errors presented and discussed in medicine conferences,
6 contained errors actually occurring on the medicine service. In surgery,
71 of the 80 cases with errors that were presented and discussed actually
occurred on the surgery service.
Medicine conference leaders acknowledged having made an error during
a discussion of a case with an error somewhat less often than did surgery
conference leaders (1/18 cases vs 24/98 cases, respectively; P = .12).
Errors were attributed to an individual, team, or system less often
in medicine conferences (8/21 errors [38%]) than in surgery conferences (88/112
errors [79%]; P<.001). In medicine, 7 (88%) of
8 error attributions included the team or the system and only 1 was thought
to be due solely to an individual (Table
4). In surgery, 56 (64%) of 88 included the team or the system and
one third were thought to be due solely to an individual. This difference
was not statistically significant (P = .17).
We found differences between internal medicine and surgery conferences
in format, frequency of adverse events and errors in cases presented, discussion
of adverse events and errors resulting in an adverse event or patient death,
and attribution of errors to a particular cause. Internal medicine conferences
were lecture based, with relatively little discussion. Cases with adverse
events and errors were infrequently presented and, when presented, were discussed
as an error only half of the time. Thus, a resident attending 48 medicine
morbidity and mortality conferences in a year would observe only 7 error discussions
on average, since each conference presented an average of 1.5 cases and an
error was discussed in 10% of cases. Errors were attributed to a cause in
less than half of the cases, with an emphasis on team and system causes of
Surgery conferences focused on case presentation and audience discussion.
Most cases presented had an adverse event that was often associated with error.
When a case with an adverse event and error was presented, it was almost always
discussed as an error. Thus, a resident attending 48 surgery morbidity and
mortality conferences in a year would observe 44 error discussions on average,
since each conference presented an average of 2.7 cases and an error was discussed
in 34% of cases. Errors were attributed to a particular cause, with one third
of errors attributed solely to an individual. In both medicine and surgery
conferences, errors were infrequently discussed explicitly, and leaders infrequently
acknowledged ever having made an error.
Our findings contrast with the beliefs of internal medicine residency
directors, as recently reported by Orlander et al.7 More
than 80% of 295 responding internal medicine program directors reported that
morbidity and mortality "cases were most often selected because of unexpected
adverse events or suspected error" and "that when present, medical error was
discussed with moderate to high success". This may reflect the difference
between desired rather than actual performance in presenting and discussing
errors. Our results are, however, consistent with previous studies of self-reported
errors11- 13 and
sociological studies of internal medicine residents14,15 and
Why is it important that adverse events and errors are presented and
discussed? Discussion of errors with the goal of learning how to prevent them
underlies the tradition of the morbidity and mortality conference7,8,17 and is supported by
principles of adult learning.18- 20 Open
discussion of errors may enhance error reporting21 and,
thus, promote patient safety. Increased error reporting has led to safety
improvements in other industries and is promoted by leaders in the fields
of medicine and safety.1 Furthermore, as systems
of care become targets for improvement, increased attention to physician competency
in assessing system contribution to errors becomes more important. In fact,
morbidity and mortality conferences are unrealized opportunities to provide
residents with experiences that help develop competency in all 6 core areas
required by the Accreditation Council for Graduate Medical Education (ACGME),
including systems-based practice, practice-based learning and improvement,
professionalism, and communication.22 Modeling
error disclosure in a supportive, nonblaming environment may benefit residents
by training them for successful personal management of this intrinsic challenge
of medical practice.7,12,13,23- 25 As
a profession, physicians are entrusted with monitoring and improving the quality
of the medical care they provide.26,27 This
responsibility includes identifying and remedying those services and procedures
that threaten patient safety. Error discussion is a vital aspect of fulfilling
The paucity of adverse event and error presentation and discussion in
internal medicine compared with surgery is likely related to regulatory and
cultural factors. The ACGME,22 the only national
organization that addresses whether a morbidity and mortality conference should
occur and what should take place during the conference, requires that surgery
morbidity and mortality conferences present and discuss "all deaths and complications
that occur on a weekly basis." There is no similar requirement for internal
medicine. Without a specific requirement to do so, adverse events and errors
occurring in the medicine service are not generally discussed. Moreover, our
findings support a preference to present and discuss errors occurring in other
services. The culture of internal medicine still "seems to have no place for
Our findings demonstrate important cultural differences between internal
medicine and surgery and missed opportunities for learning to improve patient
care. For departments of medicine, efforts to ensure that adverse events and
errors are presented and discussed with a view to improving systems and enhancing
disclosure may be especially valuable. Departments of surgery are fulfilling
their mandate to present and discuss adverse events and errors but may overemphasize
the role of the individual and underemphasize underlying system defects. In
both departments, conference leaders have the opportunity to model error acknowledgment
and use explicit language in error discussion more frequently and ensure that
efforts are clearly linked to education and local improvement activities.