R E S E A R C H R E P O R T

Assessing Medical Students Training in End-of-life

Communication: A Survey of Interns at One Urban
Teaching Hospital
Wayne A. Ury, MD, Cathy S. Berkman, PhD, Catherine M. Weber, PhD, Monica G. Pignotti, MSW,
and Rosanne M. Leipzig, MD, PhD

ABSTRACT

Purpose. Although interns are responsible for caring
for dying patients, little is known about end-of-life
education and training, including communication skills,
in U.S. medical schools. This study of three consecutive
cohorts of new interns assessed their perceptions of the
amount and types of classroom and clinical instructional
strategies used during medical school, their self-rated
skill and comfort levels in different aspects of end-of-life
communication, and the associations between these
measures.
Method. A self-administered questionnaire was given
to three consecutive cohorts (19961998) of incoming
interns (n 162). Measures were self-reported amount
and type of education and clinical experience with four
end-of-life communication domains (giving bad news,
discussing advance directives, discussing prognosis with
the patient, and discussing with the patients family) and
self-perceived comfort and skill levels in relation to
different types of end-of-life communication.
Results. A total of 157 interns completed the
questionnaire. They reported very little classroom
teaching, clinical observation, or clinical experience
with end-of-life communication during medical school.
They lacked comfort and skill in the end-of-life
communication domains that were studied. More
reported clinical observation and experience with caring
for and communicating with dying patients was associ-
ated with greater perceived comfort and skill, while
classroom teaching was not.
Conclusions. These interns, mostly U.S. medical
school graduates (98.7%, n 155) reported little
training and low self-perceived comfort and skill with
important elements of end-of-life communication that
might contribute to a lack of preparedness to address
these issues during their internship. Further research
that confirms and explains the underlying reasons for
these findings seems warranted.
Acad. Med. 2003;78:530537.

First-year internal medicine residents
(interns) are expected to assume sig-
Dr. Ury is assistant professor of medicine, New York
Medical College, Valhalla, New York, and chief,
Section of Palliative Medicine, Saint Vincents
Catholic Medical Center of New York, New York.
Dr. Berkman is associate professor, Fordham
University Graduate School of Social Service, New
York, New York. Dr. Weber is associate professor of
family medicine and assistant dean for curriculum,
University of Connecticut School of Medicine. Ms.
Pignotti is a research associate of the Sections of
Palliative and Geriatric Medicine, Saint Vincents
Catholic Medical Center, New York, New York. Dr.
nificant responsibility for patients un-
der their care, including being able to
Leipzig is professor of Medicine and vice chairperson
of the Department of Geriatric Medicine, Mount
Sinai School of Medicine of New York University,
New York, New York.
Correspondence and requests for reprints should be
addressed to Dr. Ury, Section of Palliative Medicine,
Department of Medicine, Saint Vincents Catholic
Medical Center of NY, 153 West 11 Street, Room
1213, New York, NY 10011; telephone: (212) 604-
7312; fax (212) 604-2128; e-mail: hwury@
aol.comi.
effectively communicate with patients
and their family members about end-
of-life care issues.13 Interns are often
charged with delivering bad news,
discussing prognoses, and addressing
advance care planning and surrogate
decision making, including health care
proxies and do-not-resuscitate (DNR)
orders. The many complex and in-
dividualized care decisions that must
be made in the context of difficult
psychological and social circumstances
make end-of-life communication a

530 ACADEMIC MEDICINE, VOL. 78, NO. 5 / MAY 2003

difficult task, but it is one that is
critical to good care for the patient and
the patients family.15
Poor end-of-life communication is all
too common4,6 and has been shown to
be associated with bad patient-care
outcomes, including inadequate pain
and symptom relief,7,8 delivery of un-
wanted care,4,9 and conflicts between
physicians and surrogates over deci-
sions that need to be made.6,9
Since interns are often responsible for
talking with patients and patients
families about end-of-life issues, they
need to have attained some basic level
of skill and confidence in these types of
communication by the completion of
medical school. Yet, there are few stu-
dies of how and what medical students
are taught about end-of-life communi-
cation, and it is unclear whether their
education and training prepare them for
this important role.10,11 There are
limited data on the proportion of U.S.
medical students who are educated
about end-of-life communication and
limited data on the timing and amount
of teaching,1214 and very little is known
about the teachings actual and per-
ceived content,10,11,15,16 or about med-
ical students and graduates perceptions
of its effectiveness11,1518 and the ade-
quacy of clinical exposures, including
modeling by senior physicians.16,18,19
We designed our survey of three con-
secutive cohorts (19961998) of new
interns to measure their perceptions
of the timing, instructional methods,
amount, and content of undergraduate
medical school education and clinical
training in end-of-life communication;
assess their self-rated skills and comfort
levels in communicating with patients
at the end of life; and examine the
association between these measures.
M ETHOD
Sample
All beginning internal medicine resi-
dency (three year categorical, one year
preliminary) interns in July 1996, 1997,
and 1998 at Saint Vincents Catholic
Medical Center, a 500-bed private
teaching hospital in New York City,
were invited to participate in our study.
A total of 157 of the 162 eligible
interns completed the self-adminis-
tered questionnaire during their orien-
tations, a response rate of 96.9%.
Measures
We used the following methods to
identify the domains to be included in
the questionnaire and to strengthen
content validity: focus groups of med-
ical interns, chief residents, and nurses;
interviews with experts in pain man-
agement, survey research, and medical
education; and individual interviews
with interns, residents, and attending
physicians with palliative care experi-
ence. We identified four end-of-life
communication domains (giving bad
news, discussing advance directives,
discussing prognosis with the patient
or family, and discussing end-of-life
care issues with the patients family),
and then developed measures of learn-
ers recall of their undergraduate med-
ical education teaching and clinical
experiences, and self-perceived skill
and comfort.
The questionnaire was pre-tested in
the cohort of interns entering in July
1995, and revised based on item
analysis, three focus groups, cognitive
interviewing20 of those who com-
pleted the questionnaire, and feed-
back from consultants in survey
research and medical education. Each
revision was then pre-tested on grad-
uating fourth-year medical students
using cognitive interviewing after
administration to identify problematic
items. As a result of the pre-testing,
we added five new items in 1997 and
1998, and two were eliminated. Items
added in 1997 or 1998 are indicated
in the Results.
Reported undergraduate medical
education experiences. We assessed
formal educational instruction about
end-of-life care by asking whether the
participant recalled having classroom
teaching (lectures, seminars, role play-
ing, or experience with standardized
patients about each of the four com-
munication domains being studied, and
whether such teaching had occurred
during the pre-clinical (first two) years
of medical school, clinical (last two)
years, both, or neither. We measured
the interns recall of having observed
physicians involved in these types of
communication by asking how many
times during medical school (0, 13,
46, 710, .10) they had observed
a physician in the clinical setting
having discussions in each of the end-
of-life communication domains. Interns
were also asked how many times they
had observed the withdrawal of a life-
sustaining ventilator, because this type
of end-of-life care involves good com-
munication but is not a communication
skill in and of itself.
We measured the interns recall of
direct clinical experience caring for
a dying patient, using a morphine drip
to manage a patients pain, and caring
for a patient for whom artificial nutri-
tion or hydration were ethical issues by
asking the number of times (0, 13, 4
6, 710, .10) during clinical rotations
they had participated in each of these
types of clinical care.
Self-rated comfort level. The in-
terns self-rated comfort levels in giving
bad news, discussing advance direc-
tives, diagnosing major depression,
treating major depression, and manag-
ing hypertension and congestive heart
failure (CHF) were assessed using
Likert-type response categories (low
to excellent). We included comfort
with diagnosis and treatment of major
depression for comparison with comfort
levels for giving bad news and discus-
sing advance directives because, similar
to the end-of-life communication do-
mains, the diagnosis and management
of major depression involves commu-
nication about difficult emotional

ACADEMIC MEDICINE, VOL. 78, NO. 5 / MAY 2003 531

issues that make most physicians un-
comfortable.21 We included the items
on CHF to assess interns perceived
comfort with medical management of
chronic medical conditions that are
life-threatening and eventually fatal
(CHF), and those that are not life-
threatening (hypertension), but these
items are not linked to communication
about difficult topics or to communi-
cation skills.
Self-rated skill in end-of-life care
and communication. Using four-point
Likert-type response categories ranging
from poor to excellent, we asked
interns to rate their skill levels in
caring for a dying patient, communi-
cating with a dying patient, discussing
prognosis in metastatic cancer, and
discussing prognosis for a treatable
condition.
Personal characteristics and expe-
riences with end-of life care. We
asked for each interns age, gender,
race, religion, importance of religion in
their personal life relative to medical
care, undergraduate major, and career
choice. We measured personal experi-
ence with end-of-life issues by asking
whether the interns had ever had
a family member who had been a hos-
pice patient, experienced the death of
a first-degree relative, had a life-threat-
ening illness themselves, or completed
a living will.
Data Analysis
We tested hypotheses on the associa-
tion between nominal and ordinal
variables using the chi-square statistic.
We used the Spearman correlation
coefficient to estimate correlations
between interval- and ordinal-level
variables. We used the gamma statis-
tic to estimate correlations between
two ordinal-level variables. Unless
otherwise stated, analyses were based
on the combined data from the 1996,
1997, and 1998 samples. We used
a standard statistical software for
analyzing data.
532 ACADEMIC MEDICINE, VOL. 78, NO. 5 / MAY 2003
R ESULTS
Respondents
Almost all respondents (98.7%, n
155) were graduates of U.S. medical (n
151) or osteopathic (n 4) schools.
The two remaining interns had gradu-
ated from foreign medical schools.
Respondents represented 47 U.S. med-
ical schools. The majority of the
respondents (80.9%) had attended
medical schools in the northeastern
United States; 58.6% were men, 82.8%
were aged 30 or younger, and 2.5%
were aged 40 or older. None of the
demographics or personal experiences
with items we studied was found to
differ among the cohorts, or to be
associated with any of the communi-
cation measures.
Formal Teaching and
Clinical Observation
The frequencies of formal teaching and
clinical observation for the four end-of-
life communication measures are pre-
sented in Table 1. Over 40% of the
respondents recalled having no formal
teaching of how to discuss prognosis,
give bad news, or counsel patients
families. Very few (9%26%) recalled
having had formal teaching about these
three topics during both the pre-
clinical and the clinical years. In
contrast, with regard to advance direc-
tives, 89.7% recalled having had some
formal teaching, and a greater percent-
age recalled teaching in both the pre-
clinical and clinical years when com-
pared with similar questions about
other communication domains.
Most respondents recalled observing
a physician, on three or more occasions
(see Table 1), having a discussion with
a patients family about end-of-life
issues (84.2%, 1998 sample only) or
discussing prognosis (66.7%, 1997 sam-
ple only). Approximately 50% reported
observing a physician give bad news at
least four times, and 37.8% observed
a physician discussing advance direc-
tives at least four times. Almost 90%
had had formal teaching about advance
directives, 16.7% reported never hav-
ing observed a physician discuss ad-
vance directives with a patient.
Our respondents reported little di-
rect clinical experience communicating
with or caring for dying patients during
medical school; 2.5% reported they had
never cared for a dying patient and
38.2% reported caring for one to three
dying patients. Sixty percent reported
that they had either never used or used
a morphine drip in one to three
patients to manage a patients pain
(1997 and 1998 samples only), while
36.9% reported caring for either zero or
one to three patients for whom artifi-
cial nutrition and hydration posed an
ethical dilemma. In contrast, 64%
reported having given bad news to four
or more patients.
Comfort and Skill Levels
Most respondents rated their skills in
communicating with a dying patient as
either poor (8.4%) or fair (47.4%) (see
Table 2). A greater percentage (73.8%)
rated their skills in communicating
prognosis for metastatic cancer patients
as poor or fair, compared with commu-
nicating prognosis when the patient
had a treatable condition (14.7%)
(gamma .42, p .16). Over two
thirds rated their skills in caring for
a dying patient as poor (16.1%) or fair
(52.5%).
Most respondents gave ratings of
very low, low, or fair for their comfort
levels with giving bad news (77.6%)
and discussing advance directives
(78.7%). By comparison, ratings of very
low, low, or fair were given by 47.4%
for diagnosing, by 25.6% for treating
major depression, by 47.4% for manag-
ing hypertension (1998 sample only),
and by 63.1% for managing CHF (1998
sample only).
Respondents who gave higher rat-
ings of comfort for giving bad news
Table 1
Formal Teaching and Clinical Observation Experiences with Types of End-of-life Communication of 157 Interns, Saint Vincents Catholic
Medical Center, New York, New York, 19961998
Discussion of Prognosis
Formal teaching
No formal teaching
Preclinical years only
Clinical years only
Formal teaching in both
preclinical and clinical years
Clinical observation
0 occasions
13 occasions
.3 occasions
*1997 and 1998 samples.
y1997 sample only.
z1998 sample only.
were also more likely to give higher
ratings for their comfort levels in
diagnosing (gamma .42, p .029)
or treating (gamma .61, p , .001)
major depression. There was a similar
pattern in the association between
comfort level in discussing advance
directives with diagnosing and treating
depression, although these relation-
ships were not statistically significant.
We found no association between
comfort in managing CHF or hyper-
tension and any of the communication
or clinical end-of-life measures.
Relationships between Measures
There was no significant association
between the amount or timing of
formal teaching and any of the meas-
ures of self-perceived skill or comfort.
In general, greater reported observa-
tion of end-of-life communication was
associated with higher perceived skill
and comfort levels in the four end-of-
life communication domains studied,
ACADEMIC MEDICINE, VOL. 78, NO. 5 / MAY 2003
Type of Communication
Giving Bad News
with Patient to Patient
No. (%) No. (%)
44 (42.3)* 71 (45.2)
23 (22.1) 37 (23.6)
10 (9.6) 34 (21.7)
27 (26.0) 15 (9.6)
2 (4.2)y 10 (6.4)
14 (29.1)y 68 (43.3)
32 (66.7)y 79 (50.3)
although not with greater self-per-
ceived ability to communicate with
a dying patient. Observation of the
withdrawal of a life-sustaining ventila-
tor was not associated with any of the
communication comfort or skill meas-
ures. Reported clinical experience car-
ing for and communicating with dying
patients was associated with an interns
level of comfort and skill in end-of-life
communication (see Table 2), but not
with self-perceived ability to care for
dying patients (with the exception of
number of times a student had given
bad news [rs .29, p .04]).
Self-reported skill in caring for
a dying patient was associated with
the measures of comfort and skill with
end-of-life communication. Greater
self-perceived skill in caring for a dying
patient was associated with greater
perceived skill in communicating with
a dying patient (gamma .68, p ,
.001) and discussing prognosis with
a patient with metastatic cancer
(gamma .71, p , .001), and greater
comfort in giving bad news (gamma
Discussions with Advance Directives
Patients Family Discussions with Patient
No. (%) No. (%)
56 (53.8)* 16 (10.3)
8 (7.7) 27 (17.3)
22 (21.2) 60 (38.5)
18 (17.3) 53 (34.0)
0 (0)z 26 (16.7)
9 (15.8)z 71 (45.5)
48 (84.2)z 59 (37.8)
.89, p , .001) and discussing advance
directives (gamma .67, p , .001).
Greater perceived skill in communicat-
ing prognosis for a treatable condition
was associated with greater comfort in
giving bad news (gamma .45, p
.003) and in discussing advance direc-
tives (gamma .42, p .004), but not
with perceived skill in caring for a dying
patient (gamma .11, p .25), or
communicating with dying (gamma
.34, p .063) or metastatic cancer
(gamma .21, p .16) patients.
D ISCUSSION
In our study of three cohorts of interns,
we found that (1) they reported very
little classroom teaching, clinical ob-
servation, or clinical experience with
end-of-life communication; (2) they
lacked comfort and skill in many
aspects of end-of-life communication;
and (3) the amount of reported clinical
observation and experience was associ-
ated with level of comfort and skill in
533
Table 2

Interns Self-rated Ability and Comfort with End-of-life Communication by the Number of Occasions They Were Involved in Various
End-of-life Scenarios, St. Vincents Catholic Medical Center, New York, New York, 19961998

Number of Occasions a Student was Involved in

Giving Bad News Taking Care of Using Morphine Drip to Caring for a Patient For Whom Artificial
to Patient Dying Patient Manage Patients Pain Nutrition and Hydration were Ethical Issues
03 .3 03 .3 03 .3 03 .3
Occasions Occasions Occasions Occasions Occasions Occasions Occasions Occasions
No. (%) No. (%) No. (%) No. (%) No. (%) No. (%) No. (%) No. (%)
Ability to communicate
with a dying patient
Poor/fair 17 (70.8)* 35 (64.8)* 1 (100) 52 (55.3) 21 (67.7) 33 (50.8) 23 (59.0) 30 (53.6)
Good/excellent 7 (29.2)* 19 (35.2)* 0 (0) 42 (44.7) 10 (32.3) 32 (49.2) 16 (41.0) 26 (46.4)
Ability to discuss prognosis
in metastatic cancer
Poor/fair 16 (88.9)* 43 (69.3)* 2 (100) 57 (72.2) 22 (81.5) 37 (68.5) 30 (88.2)* 29 (61.7)*
Good/excellent 2 (11.1)* 19 (30.7)* 0 (0) 22 (27.8) 5 (18.5) 17 (31.5) 4 (11.8)* 18 (38.3)*
Comfort giving bad news
to a patient
Very low/low 27 (48.2)z 24 (24.0)z 2 (50.0)* 28 (46.7)* 29 (46.0)y 22 (23.4)y 27 (46.6)* 24 (24.2)*
Fair 22 (39.3)z 48 (48.0)z 2 (50.0)* 23 (38.3)* 24 (38.1)y 47 (50.0)y 22 (37.9)* 49 (49.5)*
Good/excellent 7 (12.5)z 28 (28.0)z 0 (0)* 9 (15.0)* 10 (15.9)y 25 (26.6)y 9 (15.5)* 26 (26.3)*
Comfort discussing advance
directives with a patient
Very low/low 28 (50.0)* 35 (35.4)* 3 (75.0) 60 (39.5) 33 (52.4) 30 (32.2) 30 (51.7)y 33 (33.7)y
Fair 19 (33.9)* 41 (41.4)* 1 (25.0) 59 (38.8) 18 (28.6) 42 (45.2) 17 (50.0)y 40 (40.8)y
Good/excellent 9 (16.1)* 23 (23.2)* 0 (0) 33 (21.7) 12 (19.0) 21 (22.6) 8 (23.5)y 25 (25.5)y
*p # .05; yp # .01; zp , .00.

end-of-life communication, while class-
room teaching was not.
Our findings provide further evi-
dence that U.S. medical students may
be receiving inadequate training in
end-of-life communication skills. Based
on a review of the literature and
educational grants, Billings and Block10
concluded that end-of-life teaching is
mostly elective, predominantly only in
the preclinical years, focused on knowl-
edge rather than on attitudes and
skills, and lacking enough adequately
trained faculty who can also serve as
role models. Since interns are on the
frontlines of inpatient medical care,
they need to be somewhat prepared to
assume these responsibilities before
they begin their internships. Therefore,
this is an issue that medical schools
need to formally address.10
Even if our study underestimated the
actual amount of end-of-life communi-
cation education and clinical training
these recent graduates received, it did
reveal that they perceived they had
had little training in this area. Educa-
tion about advance directives is the
one exception, possibly because discus-
sion of advance directives is linked to
signing a legal document and a tangible
clinical task (i.e., whether to perform
CPR) and is, therefore, seen as an
aspect of communication that needs to
be taught.
Our findings about whether there is
formal teaching about end-of-life com-
munication, and if so, when it is
provided, are consistent with those of
previous studies. In one survey of U.S.
medical schools, 89% offered some type
of formal teaching, but only 11%
offered a dedicated course on end-of-
life care.22 Lectures were the primary
teaching modality in 83% of the
courses, and less than a third offered
sessions with patients. In a 1991 study
of medical school class presidents, over
a fourth of the respondents reported
one hour or less of class time on end-
of-life education, and 37% rated the
quality of teaching as ineffective.23
Despite an apparent lack of educa-
tion about these matters, medical
students have given favorable reviews
to education about end-of-life care2428
and have reported that these educa-
tional experiences enhanced their
comfort levels28,29 and reduced their
sense of hopelessness and anxiety3032
in caring for patients with advanced
cancer. However, because our findings
are specifically about communication

534 ACADEMIC MEDICINE, VOL. 78, NO. 5 / MAY 2003

with dying patients, they may not be
comparable with these reports about
more general end-of-life care educa-
tional experiences.
We are unaware of other studies that
have documented whether medical
students had the opportunity to observe
physicians in end-of-life communication
situations or whether they had direct
clinical experience with dying patients.
The lack of reported observation of and
participation in end-of-life communica-
tion and clinical care may reflect a lack
of comfort and skill among medical
school faculty, or that they view these
tasks as not part of their role as edu-
cators or feel patients and families will
find the presence of observers intrusive.
Further research that addresses these
issues would seem warranted.10
It is of particular concern that 64%
of the interns in our study had given
bad news to four or more patients, even
though only about 50% had recalled
observing a physician doing this on four
or more occasions. From an ethical and
legal perspective, and because of the
serious nature and the complexity of
information being conveyed, giving bad
news would seem to be a physicians,
and not a medical students, responsi-
bility.1,10 However, this task may be
delegated to students because it is
viewed as less important, a task that
does not require skill, or because the
medical student has spent more time
with the patient than the housestaff or
attending. Students may have reported
a lack of observational opportunities
because most medical school faculty
lack the requisite training to feel
capable of teaching these skills.
The concept of comfort may in-
corporate constructs such as knowledge
and skill.33 Although the term con-
fidence is used more frequently than
comfort, comfort has been used in
previous end-of-life and communica-
tion skills studies34,35 and may be a
better term because it can incorporate
confidence, attitudes, and motivation
about a clinical skill. Therefore, com-
ACADEMIC MEDICINE, VOL. 78, NO. 5 / MAY 2003
fort reflects both the clinicians general
perception and his or her confidence.
The relatively low level of comfort in
discussing bad news and advance
directives most likely reflects both the
complexity of the issues and a lack of
adequate preparation during medical
school. Nonetheless, the high propor-
tion of our respondents who reported
low levels of comfort would seem to
indicate that some medical graduates
have not been adequately prepared to
communicate with dying patients.
Despite research showing that most
primary care physicians have difficulty
diagnosing and treating depression,21
the interns in our study reported
significantly greater comfort with these
practices than with discussing advance
directives or giving bad news. This may
be due to an inflated view of their skills
in relation to depression as well as a lack
of skills in end-of-life communication.
Personal experience with end-of-life
issues, religion, undergraduate major,
and career choice were not associated
with the communication measures,
possibly reflecting a lack of personal
experience with death and dying, as
well as the relatively young age of most
of the interns. However, the small
number of interns in our study who
had had personal experiences with
end-of-life care issues made it difficult
to speculate about this relationship.
Strengths and Limitations
The strengths of our study were a high
response rate; the large number of
medical schools (n 47) represented
by the respondents; multiple measures
of the amounts, types, and timings of
medical education and training; and
multiple measures of comfort and skill
in end-of-life communication.
Our study also had limitations. The
sample was small and drawn from only
one hospital. The amount and timing
of education and clinical experience
were measured retrospectively, based
on self-report, and may be subject to
recall bias. The possibility of reverse
causality cannot be ruled out in this
cross-sectional design. The three items
measuring skill and the four items
measuring comfort do not adequately
assess comfort or skill in end-of-life
communication. However, ours is the
only study that has measured both
multiple aspects of undergraduate med-
ical education (classroom and clinical
teaching) and comfort and skill level in
end-of-life communication.
The validity of the items we used to
measure experience and self-perception
has not been established, although
similar questions have been used in
previous questionnaires.18,36,37 Al-
though the focus group participants
and individual interviews consistently
conceptualized the term comfort, we
did not define what we meant by
comfort in the questionnaire, and the
interns may have interpreted it in
various ways. We also did not assess
actual knowledge or skill in end-of-life
communication with an objective test
or a performance assessment such as an
OSCE or a standardized patient assess-
ment, and, therefore, we cannot equate
a respondents self-rating of comfort or
skill with actual ability. Despite these
concerns, the absence of significant
differences on the study measures
among the three cohorts of interns in
our study lends confidence to the
reliability of these measures.
In addition to replicating these
findings with larger samples, more work
is needed to develop reliable and valid
measures. More objective and specific
measures of the classroom education
and clinical experience are needed to
evaluate how the amounts and types of
educational and clinical experiences
are associated with knowledge, atti-
tudes, skills, and comfort levels. More
specific measures of knowledge and
skill levels are also needed to assess
whether medical students are properly
trained to provide end-of-life care.
Longitudinal studies following med-
ical students from the start of under-
535
graduate medical education through
residency training would provide valu-
able information about the causal re-
lationships among education (amount,
timing, and modalities) and knowledge,
attitudes, skills, and comfort with end-
of-life communication. The role of men-
toring should be assessed, because it has
been shown that faculty role models are
associated with better attitudes regard-
ing other aspects of end-of-life care and
improved self-perception.16,18,19
The findings of our study suggest
deficiencies in formal education and
clinical observation and experience
during medical school, as well as low
levels of self-perceived skill and com-
fort with different aspects of end-of-life
communication. While greater reported
observation of the end-of-life commu-
nication scenarios presented was asso-
ciated with greater perceived skill and
comfort, classroom teaching and clini-
cal experiences (other than those that
are specifically end-of-life communica-
tion) did not appear to contribute to
a greater sense of preparedness.
Our findings also suggest that end-of-
life communication is a clinical skill that
requires specific and greater attention
in medical school curricula. A single
lecture about the psychological issues
experienced by terminally ill patients or
a few clinical experiences using a mor-
phine drip to manage a patients pain
are not sufficient training in end-of-life
communication. Although more class-
room education and clinical training
about end-of-life communication in
medical school seems warranted, further
research that clarifies the core educa-
tional issues and the most effective
instructional strategies is needed.
R EFERENCES
1. Ong L, deHaes J, Lammes F. Doctorpatient
communication: a review of the literature.
Soc Sci. 1995;40:90318.
2. Virmani J, Scheiderman L, Kaplan R. Re-
lationship of advance directives to physician
patient communication. Arch Intern Med.
1994;154:90913.
536 ACADEMIC MEDICINE, VOL. 78, NO. 5 / MAY 2003
3. Hofmann J, Wenger N, Davis R, et al.
Patient preferences for communication with
physicians about end-of-life decisions. Ann
Intern Med. 1997;127:112.
4. Hakim R, Teno J, Harrell F, et al. Factors
associated with donotresuscitate orders:
patients preferences, prognoses and physi-
cians judgments. Ann Intern Med. 1996;
125:28493.
5. Morasso G, Capelli M, Viterbori P, et al.
Psychological and symptom distress in termi-
nal cancer patients with met and unmet
needs. J Pain Symptom Manage. 1999;17:
4029.
6. SUPPORT Principal Investigators. A con-
trolled trial to improve care for seriously ill
hospitalized patients: the study to under-
stand prognoses and preferences for out-
comes and risks of treatments (SUPPORT).
JAMA. 1995;274:15918.
7. Ingham J, Foley K. Pain and the barriers to
its relief at the end of life: a lesson for
improving end of life health care. Hospice J.
1998;13:89100.
8. Grossman S, Sheidler V. Skills of medical
students and house officers in prescribing
narcotic medications. J Med Educ. 1985;60:
5527.
9. Teno J, Lynn J, Wenger N, et al. Advance
directives for seriously ill hospitalized patients:
effectiveness with the patient self-deter-
mination act and the SUPPORT intervention.
J Am Geriatr Soc. 1997;5:5007.
10. Billings J, Block S. Palliative care in un-
dergraduate medical education. JAMA.
1997;278:7338.
11. Buss M, Marx E, Sulmasy D. The prepare-
dness of students to discuss end-of-life issues
with patients. Acad Med. 1998;73:41822.
12. Vetto J, Elder N, Toffler W, Fields S.
Teaching medical students to give bad news:
does formal instruction help? J Cancer Educ.
1999;14:137.
13. Barnard D, Quill T, Hafferty F, et al. Pre-
paring the ground: contribution of preclin-
ical years to medical education of care near
end of life. Acad Med. 1999;74:499505.
14. Ogle KS, Mavis B, Rohrer J. Graduating
medical students competencies and educa-
tional experiences in palliative care. J Pain
Symptom Manage. 1997;14:2749.
15. Kaye J, Gracely E, Loscalzo G. Changes in
students attitudes following a course on
death and dying: a controlled comparison.
J Cancer Educ. 1994;9:7781.
16. Weissman D, Griffie J, Gordon D, Dahl J. A
role model program to promote institutional
changes for management of acute and can-
cer pain. J Pain Symptom Manage. 1997;14:
2749.
17. Hayes R, Stoudemire A, Kinlaw K, Dell M,
Loomis A. Changing attitudes about end-
of-life decision making of medical students
during third-year clinical clerkships. Psycho-
somatics. 1999;40:20511.
18. Caralis P, Hammond JS. Attitudes of medical
students, housestaff, and faculty physicians
toward euthanasia and termination of life-
sustaining treatment. Crit Care Med.
1992;20:68390.
19. Weissman DE, Dahl JL. Update on the
cancer pain role model education program.
J Pain Symptom Manage. 1995;10:2927.
20. Jobe J, Mingay D. Cognitive research im-
proves questionnaires. Am J Public Health.
1989;79:10535.
21. Badger LW, deGruy FV, Hartman J, et al.
Patient presentation, interview content, and
the detection of depression by primary care
physicians. Psychosomat Med. 1994;56:
116470.
22. Holleman W, Holleman M, Gershenhorn S.
Death education curricula in U.S. medical
schools. Teach Learn Med. 1994;6:2603.
23. Merman A, Gunn D, Dickinson G. Learning
to care for the dying: a survey of medical
schools and a model course. Acad Med.
1991;66:358.
24. Barton D, Flexner J, vanEys J, Scott C.
Death and dying: a course for medical
students. J Med Educ. 1972;47:94551.
25. Kitchen A. Potentials and pitfalls in death
education in medical schools. In: DeBellis R,
Marcus E, Kutscher A, Klagsbrun S, Seeland
I Preven (eds). Thanatology Curriculum
Medicine. New York: Hawarth Press, 1988:
313.
26. Goedell B, Donahue J, Benoiel JQ.
Death education in medical school: a semi-
nar on terminal illness. Death Educ. 1982;
8:3639.
27. Martin RI, Wylie N. Teaching third-year
medical students how to care for terminally
ill patients. Acad Med. 1989;64:4134.
28. Weissman D. Pre-clinical palliative medicine
education at the Medical College of Wis-
consin. J Cancer Educ. 1993;8:1915.
29. Anderson J. Evaluation of a practical ap-
proach to teaching about communication
with terminal cancer patients. Med Educ.
1982;16:3027.
30. Cassileth B, Egen T. A modification of
medical student perception of the cancer
experience. J Med Educ. 1979;54:797802.
31. Blanchard C, Ruckdeshel J, Cohen R, Shaw
E, McSharty J, Horton J. Attitudes towards
cancer: the impact of a comprehensive
oncology course on second-year medical
students. Cancer. 1981;47:275662.
32. Kaye JM. Will a course on death and dying
lower students death and dying anxiety?
J Cancer Educ. 1991;6:214.
33. Lurie N, Margolis K, McGovern P, Mink P.
Physician self-report of comfort and skill in
 providing preventive care to patients of
the opposite sex. Arch Fam Med. 1998;7:
1347.
34. Weissman DE, Ambuel B, Norton AJ,
Wang-Cheng R, Schiedermayer D. A survey
of competencies and concerns in end-of-life
care for physician trainees. J Pain Symptom
Manage. 1998;15:8290.
35. Haas D, Coe R. Do physicians discuss HIV
and AIDS with patients? A survey of
physician practices. J Community Health.
1997;22:40116.
36. Scurry M, Bruhn J, Bunce H. The house
officer and the dying patient: attitudes. Gen
Hosp Psychiatry. 1979;1:3015.
37. Bruhn J, Bunce H, Scurry M. Texas physi-
cians attitudes toward death and dying.
Texas Med. 1982;78:626.

ACADEMIC MEDICINE, VOL. 78, NO. 5 / MAY 2003 537