JOURNAL OF PALLIATIVE MEDICINE

Volume 8, Supplement 1, 2005

© Mary Ann Liebert, Inc.

Interventions to Enhance Communication among

Patients, Providers, and Families

JAMES A. TULSKY, M.D.

ABSTRACT

Whether patient suffering is caused by physical symptoms, unwanted medical intervention,

or spiritual crisis, the common pathway to relief is through a provider who is able to elicit
these concerns and is equipped to help the patient and family address them. This paper re-
views the current state of knowledge in communication at the end of life, organized accord-
ing to a framework of information gathering, information giving, and relationship building;
and then focuses on interventions to enhance communication among patients, providers, and
families.
Several observations emerge from the existing literature. Patients have highly individual-
ized desires for information and we cannot predict patient preferences. Communication cod-
ing methodology has advanced significantly yet the current systems remain poorly under-
stood and largely inaccessible. Physicians and other health care providers do not discuss
sufficiently treatment options, quality of life or respond to emotional cues from patients, and
there is plenty of room for improvement. On the positive side, we have also learned that
physicians and other health care providers can be taught to communicate better through in-
tensive communication courses, and that communication interventions can improve some pa-
tient outcomes. Finally, huge gaps remain in our current knowledge, particularly with regard
to understanding the relationship between communication style and outcomes.
These findings suggest several recommendations. We should create larger and more diverse
datasets; improve upon the analysis of recorded communication data; increase our knowledge
about patient preferences for information; establish a stronger link between specific com-
munication behaviors and outcomes; and identify more efficient ways to teach providers com-
munication skills.

“good death” differently1; family needs at the end of life and to negotiate
E VERYONE DEFINES A
and whether patient suffering is caused by
physical symptoms, unwanted medical interven-
individually the goals of care.
Communication success and failure appears to
tion, or spiritual crisis, the common pathway to generate more gratitude and complaints than any
relief is through a provider who is able to elicit other aspect of the work performed by health care
these concerns and is equipped to help the pa- professionals,2 and poor communication has been
tient and family address them. Good communi- identified as a primary source of malpractice
cation enables providers to uncover patient and suits.3–5 Within the field of palliative care, con-

Center for Palliative Care and the Department of Medicine, Duke University, Durham, North Carolina, and the
Veterans Affairs Medical Center, Durham, North Carolina.

S-95
S-96 TULSKY

sultations are commonly requested for assistance information in an emotionally supportive way.8
with communication, including establishing Patients want to discuss emotional concerns, but
goals of care and delivering prognostic informa- frequently are unwilling to bring them up spon-
tion.6 Patients and families regularly cite com- taneously and may need to be prompted.9
munication skills as an essential component of These findings suggest that communication
end-of-life care.1 A number of cross-sectional and with patients at the end of life that should be in-
retrospective studies infer a relationship between formative, patient-centered rather than physi-
the quality of communication and patient-cen- cian-centered,10 and attentive to patients’ emo-
tered outcomes; and, anecdotally, most observers tional needs. More specific models of
recognize that how clinicians talk to patients communication have been proposed, particularly
clearly makes a difference in the quality of their for tasks such as delivering bad news,11–16 dis-
care. Yet, constructing studies to demonstrate cussing disease progression and end-of-life deci-
firmly this relationship has met many challenges. sions,17,18 and discussing palliative care.19 Al-
In particular, only a handful of randomized, con- though an overarching theory of communication
trolled trials have attempted to study the effect at the end of life has not been proposed, the most
of communication interventions on distal out- commonly advocated approaches borrow from
come measures such as mortality or quality of life. models of patient-centered communication, in-
Although the quality of communication is gener- formed consent, and Rogerian notions of uncon-
ally recognized as critically important to the clin- ditional positive regard. Studies of communica-
ical encounter, the data supporting its effects, tion behavior tend to compare current practice
both positive and negative, are slim. Thus, sig- with idealized behaviors according to these the-
nificant opportunities exist for scientific investi- oretical models, and most teaching interventions
gation in this area. have been developed to train providers to use
Communication research has progressed si- more of these behaviors.
multaneously on several fronts—discerning what This paper will briefly review the current state
elements of communication are preferred by pa- of knowledge in communication at the end of life,
tients and contribute most to their well-being, ob- organized according to a commonly used frame-
serving provider communication behaviors and work in which provider–patient communication
assessing their skill, and evaluating interventions is described as consisting of three inter-related
to improve communication. Ideally, such re- tasks: (1) information gathering, (2) information-
search would involve a sequential series of stud- giving, and (3) relationship building.20 Each of
ies, with new interventions developed and tested these three communication functions has obvious
based upon firm evidence previously derived. In- implications for clinical care. After this review, I
stead, because of the paucity of data and the per- focus discussion on the well-designed evaluation
ceived need for rapid change, many interventions studies of interventions to enhance communica-
have been created based primarily on conceptual, tion among patients, providers, and families.
rather than empirical, models of ideal communi-
cation.
These conceptual models tend to coalesce INFORMATION GATHERING
around several goals identified in survey re-
search. According to one recent study of dying Information gathering enables clinicians to as-
patients, family members and health care sess patients’ current health status and needs. It is
providers these goals include: talking with pa- the primary communication tool to facilitate diag-
tients in an honest and straightforward way, be- nosis and treatment. Studies looking at informa-
ing willing to talk about dying, giving bad news tion gathering document that both physicians and
sensitively, listening to patients, encouraging nurses, even in hospice settings, tend to underes-
questions, and being receptive to when patients timate, and commonly do not elicit, the full range
are ready to talk about death.7 Patients wish for of concerns of terminally ill patients.21 Rather than
their providers to achieve a balance between be- using facilitative communication techniques such
ing honest and straightforward and not discour- as open-ended questions or empathic responses,
aging hope. Another study found that although they often block discussion of psychosocial issues
patients must receive adequate information to by changing the subject or ignoring patients’ emo-
make informed choices, they wish to receive that tional states.22–24 As a result, patients tend not to
COMMUNICATION AMONG PATIENTS, PROVIDERS, AND FAMILIES S-97

disclose the majority of their concerns, which leads issues? A cross-sectional survey was performed
to inaccurate assessments of patient distress.25 One that included 100 consecutive patients with
large study of terminally ill patients who visited chronic obstructive pulmonary disease to assess
their oncologists found that physicians dedicated their desires to discuss end-of-life care.33 Of these
only 23% of their time discussing health-related patients, 85% wished to talk about this topic, yet
quality-of-life issues, and frequently missed op- only 21% had already had such discussions. Pa-
portunities to address issues that seemed to be tients’ preferences for discussing end-of-life is-
most important to patients.26 sues were not associated with pulmonary status,
medication use, functional status, history of hos-
pitalizations, or even experience with mechanical
INFORMATION GIVING ventilation. The authors concluded that disease
severity cannot be used to predict when to talk
Information giving is the process through to patients.
which providers educate patients and families, Similarly, ethnic differences clearly play a role
and enlist them as partners in their own care. As in communication preferences. For example, pa-
with information gathering, shortcomings exist in tients of Korean and Mexican background are less
information giving. The delivery of bad news of- likely to want to hear bad news directly and are
ten does not meet patient needs,14,16,27 and dis- more likely to defer to their families.34 Similar dis-
cussions of advance care planning are infrequent parities exist in preferences for treatment at the
and their quality falls short of expert recommen- end of life; for example, African-American pa-
dations.28–30 Just as the notions of a good death tients are less likely to complete advance di-
differ among patients, we have also learned that rectives and more likely to request aggressive
patients have highly individualized and nuanced treatment than their Caucasian and Latino coun-
desires for information. In one study, 126 Aus- terparts.35, 36 Nevertheless, one cannot predict
tralian cancer patients who were recently diag- preferences based on ethnic background alone,
nosed as having incurable disease were asked and working with a diverse population requires
about their preferences for prognostic informa- attention to cultural competence.37
tion and were also assessed for anxiety and de-
pression.31 More than 95% of these patients
wanted information about side effects, symp- RELATIONSHIP BUILDING
toms, and treatment options, and 80–85% of them
wanted to hear survival rates. In general they pre- Relationship building encompasses communi-
ferred to hear this information as words, rather cation that serves to create an environment in
than seeing graphs or charts. Of the patients, 59% which patients are more trusting, more adherent
wanted to discuss survival when they first to recommendations, and more likely to have
learned that their disease was metastatic, and 40% their emotional needs met. We know that patients
wanted to negotiate the timing of these discus- appreciate discussions about end-of-life concerns
sions. There was a lower frequency of depression and that such communication generally en-
among patients who never wanted to discuss ex- hances, rather than detracts from, the depth of
pected survival rates. provider–patient relationships.38, 39 Nevertheless,
In a study by the same group, 118 patients with numerous studies illustrate that providers do not
incurable cancer visiting nine oncologists were attend sufficiently to patient affect in these con-
audio-recorded as they received the news of their versations. In an Australian audiotape study of
situation.32 Of these patients, 85% were told the 298 patients with cancer visiting their oncologists,
aim of chemotherapy, 75% were told that the dis- the encounters were coded for the presence of pa-
ease was incurable, and 58% discussed life ex- tient cues for information or emotional support.40
pectancy. Notably, in this study, more informa- Investigators found that patients offered a me-
tion did not increase patient anxiety. However, dian of two cues per conversation and that physi-
patients who were encouraged by physicians to cians responded to 72% of the informational cues
participate in decision making experienced but only 28% of the emotional cues. Only 11% of
greater anxiety. these physicians responded to all of the emotional
Is there any way to predict what patients want cues, yet whether oncologists responded to these
to know, or when they are ready to discuss these cues had no effect on anxiety or satisfaction. Fi-
S-98 TULSKY

nally, a recent study examined audiotapes of 51 In contrast, one recent British trial has shown
family conferences in the intensive care units of the effectiveness of intensive interventions in can-
four hospitals. Investigators examined the pro- cer communication. Researchers randomly as-
portion of doctor versus family talk time and its signed 160 oncologists to receive a 3-day facili-
impact on satisfaction.41 The results indicate that tated small group communication training,
clinicians spoke 71% of the time, but increased written feedback, both training and written feed-
proportion of the conference time spent on fam- back, or no intervention.47 Encounters between
ily talk correlated with increased family satisfac- these oncologists and 2407 of their patients were
tion with the communication. These data confirm videotaped before and after the intervention. The
that patient–family–provider communication is videotapes were rated using Medical Interaction
far from ideal, and many of these studies suggest Process System (MIPS), a coding system that as-
that patients would benefit if communication sesses patient-centered communication behav-
could come closer to its ideal model. iors. Other outcomes included physicians’ psy-
chosocial beliefs and self-perceived competency.
The analysis demonstrated that physicians who
INTERVENTIONS received the communication training were more
likely to use techniques such as open-ended ques-
These data have stimulated a number of inter- tions, expressions of empathy, and appropriate
ventions to improve communication at the end of responses to patients’ cues. Intervention physi-
life. Broadly speaking, one group of interventions cians’ attitudes and beliefs toward psychosocial
has attempted to improve physician communica- issues also improved,48 and a number of the im-
tion skills directly. A second series of interven- provements in communication behaviors per-
tions has facilitated the transfer of information sisted at 12-month follow-up.49
about patient beliefs or preferences to providers, This study provides fairly conclusive evidence
or directly helped negotiate their relationships. that physicians can be taught critical communi-
In the first group, there have been numerous cation skills and subsequently modify their be-
programs to train physicians and other health care haviors. However, even if such a program could
providers in communication skills. Many have be replicated and disseminated widely (an orga-
been delivered in the primary care or hospital- nizational and financial challenge), we still do not
based setting and are not specifically focused on understand what types of communication have
communication with patients at the end of life. Al- an effect on patient-level outcomes. In this study,
though many of the reported examples have been patient satisfaction was not changed among the
small, uncontrolled studies, there are several ran- intervention physicians.50 In addition to the most
domized controlled trials of teaching interven- straightforward explanation that the communi-
tions.42, 43 In general, these have demonstrated that cation has no impact on satisfaction, there may
communication skills can be taught, and that the be alternative explanations for this finding. Pa-
most effective methods are intensive (at least sev- tients’ satisfaction ratings are often so positively
eral days in length) and use active learning with skewed at baseline (i.e., patients are often highly
role play and, frequently, standardized patients. satisfied) that it is difficult to produce a mean-
In the setting of patients with life-limiting dis- ingful change. Furthermore, the instruments
ease, short interventions that depend on didactic available to measure satisfaction are not sensitive
teaching have been ineffective. For example, one enough to detect subtle changes in patient satis-
course that used lecture, review of a videotape, faction. Other outcomes that were not measured
and one observed conversation did not dramati- in this study, such as quality of life, patient mood,
cally increase the number of advance directive or adherence to therapy, may be more important.
discussions or completions among an outpatient The second group of interventions comprises
population.44 Most of these interventions have those designed to affect directly the communica-
targeted physicians; however, some investigators tion process between clinician and patient. The
have always included nurses in clinician com- best known example was the Study to Under-
munication programs,45 and there are specific stand Prognoses and Preferences for Outcomes
curricula such as the End-of-Life Nursing Educa- and Risks of Treatment (SUPPORT), a multicen-
tion Consortium (ELNEC) aimed at nurses and ter trial in which trained nurses elicited patient
other health care professionals.46 preferences for treatment and provided these, to-
COMMUNICATION AMONG PATIENTS, PROVIDERS, AND FAMILIES S-99

gether with prognostic estimates, to the physi- prospective, multicenter randomized controlled
cians in an effort to stimulate discussion and de- trial of ethics consultation compared to usual
cision making.51 The intervention had no impact care. Seven intensive care units were involved in
on any patient outcomes, including the number the study, and investigators enrolled 551 patients
of discussions about end of life. Again, there are with value-related treatment conflicts. Outcomes
many potential explanations, ranging from the included the number of ICU days and life-sus-
possibility that patient preferences do not drive taining treatments administered to patients who
decision making, to the ineffectiveness of nurses died, the overall mortality rate, and acceptance of
in such a process. the ethics consultation by physicians, nurses, pa-
In contrast, Dutch scientists have successfully tients, and families. There was no difference in
increased patient–provider communication about mortality found between groups (63% vs. 58%).
quality-of-life issues.52 In a prospective, random- This finding is important because it addresses the
ized, controlled trial involving 214 Dutch patients potential concern that ethics (or palliative care)
and 10 oncologists, patients completed the consultations might advocate for aggressive
EORTC-30, a health-related quality-of-life sur- withdrawal of invasive treatments and actually
vey, before three successive visits. The data from lead to increased mortality. Compared with the
these visits was then provided in a graphical for- control group, hospital days (11.62 vs. 8.66; P

mat to the patients and their physicians. The en- 0.01) and ICU days (7.86 vs. 6.42; P
0.03) were
counters were audio-recorded and coded for significantly reduced, and intervention patients
health-related quality-of-life talk. Investigators spent 1.7 fewer days on the ventilator. In all, 87%
found that health-related quality-of-life issues of physicians, nurses, patients, and surrogates felt
were discussed more frequently in the interven- that the consultations were helpful.
tion group. Furthermore, the intervention physi-
cians were able to identify more patients with
moderate to severe health-related quality-of-life WHAT HAVE WE LEARNED?
problems. Perhaps most importantly, patients in
the intervention group were noted to have im- One can draw several conclusions from the ex-
proved mental health and role function, presum- isting literature on provider–patient communica-
ably because these issues were now addressed. tion at the end of life. First, patients have highly
This was a remarkably simple yet effective inter- individualized desires for information. Assump-
vention. tions about what patients do and do not want to
Others have attempted interventions that ma- hear are likely to be wrong. Therefore, at the cur-
nipulate the communication process even more rent time, one can only know a patient’s prefer-
directly. In one study, investigators facilitated a ence for information by asking.
multidisciplinary family meeting within 72 hours Second, communication-coding methodology
after patients were admitted to a critical care has significantly advanced. Many of these stud-
unit.53 Follow-up meetings assessed progress and ies depended upon the ability to collect, archive,
introduced palliative care topics as appropriate. and code a multitude of audio-recorded conver-
Patients who died in the intensive care unit used sations. The technology has advanced sufficiently
fewer critical care resources. This reduction in re- that all of this can be done digitally, with relative
source use was observed 4 years after the inter- ease, and with computer assistance.56 Further-
vention was implemented in the unit.54 The im- more, many different coding schemes exist de-
pressive results of this trial are mitigated by pending on the type of outcomes desired.57
weaknesses in its nonrandomized, time-series de- Although there are significant advantages and
sign. disadvantages to these varied approaches, the
A much more strongly designed multi-center, presence of these tools creates a sense of method-
randomized, controlled trial of ethics consulta- ological plenty where there used to be a vacuum.
tion examined the effect of facilitated communi- All of the observational studies are consistent
cation on outcomes of intensive care unit (ICU) in recognizing that physicians and other health
patients.55 Although the model was ethics con- care providers do not adequately discuss treat-
sultation, the actual intervention was quite simi- ment options or quality of life, or respond to emo-
lar to the service that is provided by palliative tional cues from patients. If we are to believe that
care consultants in such situations. This was a these communication behaviors are important,
S-100 TULSKY

significant data suggest that they are not used 2. Improve upon the analysis of recorded
sufficiently and that there is plenty of room for communication data
improvement.
Huge advances have been made in the coding
On the positive side, we have also learned that
of audio-recorded data. A number of different
physicians and other health care providers can be
coding tools now exist, but many are limited by
taught to communicate more effectively. Through
incomplete validation or a degree of specificity
intensive communication courses, providers can
that does not easily allow their use across stud-
change their behaviors and adopt a more person-
ies. The various coding tools tend to focus on dif-
centered style of communication that focuses
ferent types of analysis (e.g., content, interac-
more on the socio-emotional aspects of the en-
tional) and do not capture all of the elements of
counter. We also know that communication in-
discourse. Ideal coding methodologies would in-
terventions can improve patient outcomes, al-
tegrate diverse approaches such as content and
though the most effective interventions are
voice tone. In addition, the commonly used cod-
considerably more proactive than simply teach-
ing systems are still not broadly accessible or easy
ing communication skills.
to use. Coding manuals and scoring methods for
Finally, huge gaps exist in our current knowl-
the common systems are not simply available in
edge. We do not know what the predictors are
journal articles or by download from a website.
for patients’ preferences for information. Al-
This sometimes makes the study of communica-
though assumptions are made about the rela-
tion seem impractical to the investigator inexpe-
tionship between communication style and out-
rienced in this field. Furthermore, several soft-
comes, there are very few data supporting this
ware programs have been developed by
association. Perhaps most importantly, there are
investigators that hugely facilitate the coding and
no randomized controlled trials that change com-
archiving of communication data, but these can
munication behavior with respect to end-of-life
generally only be used in cooperation with these
care and then observe the impact of such change
investigators. Such software needs to be made
on relevant patient outcomes.
readily available, perhaps through commercial
avenues.

RECOMMENDATIONS 3. Increase our knowledge about patient

preferences for information
Recent research has clarified the nature of pro-
vider–patient communication at the end of life Much of the controversy in this field surrounds
and has rigorously demonstrated initial effec- the question of the specific pieces of information
tiveness of interventions to enhance this commu- that patients ought to receive during these en-
nication. I offer the following recommendations counters. We need a deeper understanding of pa-
for furthering this important field. tients’ preferences for information, focusing par-
ticularly on predictors for desiring certain types
1. Create larger and more diverse datasets of information. We also need to know how to as-
sess easily patients’ preferences in the clinical set-
Until very recently, most communication stud- ting and to create interventions that allow pa-
ies were quite small, analyzing between five and tients to indicate their communication needs.
50 conversations, and often with a much smaller
number of health care providers than patients. If 4. Establish a stronger link between specific
the field is to expand, it will require large analy-
communication behaviors and outcomes
ses that are valid and generalizable. Furthermore,
there will be a need to collect communication data Such a link would prove helpful in convincing
(recorded encounters) in a variety of settings in- clinicians of the value of adopting certain com-
cluding inpatient, outpatient, palliative care, and munication techniques. It would also be helpful
hospice venues. To facilitate this, collection and to have a better understanding of the optimal
archiving tools should achieve a greater degree means through which providers can communi-
of standardization, making it easy to combine cate with patients. We need to apply, wherever
datasets for analysis. In addition, we should en- possible, the methods of randomized controlled
courage more multicentered studies to observe trials to this question. Furthermore, the outcomes
and intervene on communication behavior. measured in such studies must be the most dis-
COMMUNICATION AMONG PATIENTS, PROVIDERS, AND FAMILIES S-101

tal, relevant, and patient-centered ones possible, a patient-centered approach. Psychooncology 2000;9:
such as quality of life and symptom control. 147–156.
11. Baile WF, Buckman R, Lenzi R, Glober G, Beale EA,
5. Identify more efficient ways to teach providers Kudelka AP: SPIKES—A six-step protocol for deliv-
ering bad news: Application to the patient with can-
communication skills
cer. Oncologist 2000;5:302–311.
Recent studies have provided quite convincing 12. Larson DG, Tobin DR: End-of-life conversations: evolv-
evidence that health care providers can be taught ing practice and theory. JAMA 2000;284:1573–1578.
to change and to improve their communication 13. Parle M, Maguire P, Heaven C: The development of
a training model to improve health professionals’
skills. However, all of these interventions are ex-
skills, self-efficacy and outcome expectancies when
tremely time consuming and intense, require communicating with cancer patients. Soc Sci Med
many highly trained faculty, and, as a result, are 1997;44:231–240.
expensive and not easily disseminated. We must 14. Ptacek JT, Eberhardt TL: Breaking bad news. A re-
design and test streamlined methods to teach view of the literature. JAMA 1996;276:496–502.
these communication skills to providers and see 15. von Gunten CF, Ferris FD, Emanuel LL: The patient-
whether these interventions are as effective as the physician relationship. Ensuring competency in end-
of-life care: Communication and relational skills.
current gold standards.
JAMA 2000;284:3051–3057.
16. Gillotti C, Thompson T, McNeilis K: Communicative
competence in the delivery of bad news. Soc Sci Med
REFERENCES 2002;54:1011–1023.
17. Baile WF, Glober GA, Lenzi R, Beale EA, Kudelka AP:
1. Steinhauser KE, Christakis NA, Clipp EC, McNeilly Discussing disease progression and end-of-life deci-
M, McIntyre L, Tulsky JA: Factors considered im- sions. Oncology 1999;13:1021–1031.
portant at the end of life by patients, family, physi- 18. von Gunten CF: Discussing hospice care. J Clin On-
cians, and other care providers. JAMA 2000;284: col 2002;20:1419–1424.
2476–2482. 19. Lo B, Quill T, Tulsky J: Discussing palliative care with
2. Abbott KH, Sago JG, Breen CM, Abernethy AP, Tul- patients. Ann Intern Med 1999;130:744–749.
sky JA: Families looking back: One year after discus- 20. Cohen-Cole SA: The medical interview: The three-
sion of withdrawal or withholding of life-sustaining function approach. St Louis: Mosby Year Book, 1991.
support. Crit Care Med 2001;29:197–201. 21. Maguire P, Faulkner A, Booth K, Elliott C, Hillier V:
3. Hickson GB, Clayton EW, Githens PB, Sloan FA: Fac- Helping cancer patients disclose their concerns. Eur J
tors that prompted families to file medical malprac- Cancer 1996;32A:78–81.
tice claims following perinatal injuries. JAMA 22. Ford S, Fallowfield L, Lewis S: Doctor-patient inter-
1992;267:1359–1363. actions in oncology. Soc Sci Med 1996;42:1511–1519.
4. Hickson GB, Clayton EW, Entman SS, et al: Obstetri- 23. Heaven CM, Maguire P: Disclosure of concerns by
cians’ prior malpractice experience and patients’ sat- hospice patients and their identification by nurses.
isfaction with care. JAMA 1994;272:1583–1587. Pall Med 1997;11:283–290.
5. Levinson W, Roter DL, Mullooly JP, Dull VT, Frankel 24. Maguire P: Improving communication with cancer
RM: Physician-patient communication. The relation- patients. Eur J Cancer 1999;35:1415–1422.
ship with malpractice claims among primary care 25. Ford S, Fallowfield L, Lewis S: Can oncologists detect
physicians and surgeons. JAMA 1997;277:553–559. distress in their out-patients and how satisfied are
6. Weissman DE: Consultation in palliative medicine. they with their performance during bad news con-
Arch Intern Med 1997;157:733–737. sultations? Br J Cancer 1994;70:767–770.
7. Wenrich MD, Curtis JR, Shannon SE, Carline JD, Am- 26. Detmar SB, Muller MJ, Wever LD, Schornagel JH,
brozy DM, Ramsey PG: Communicating with dying Aaronson NK: The patient-physician relationship. Pa-
patients within the spectrum of medical care from ter- tient-physician communication during outpatient
minal diagnosis to death. Arch Intern Med 2001;161: palliative treatment visits: an observational study.
868–874. JAMA 2001;285:1351–1357.
8. Parker PA, Baile WF, de Moor C, Lenzi R, Kudelka 27. Friedrichsen MJ, Strang PM, Carlsson ME: Breaking
AP, Cohen L: Breaking bad news about cancer: Pa- bad news in the transition from curative to palliative
tients’ preferences for communication. J Clin Oncol cancer care—patient’s view of the doctor giving the
2001;19:2049–2056. information. Support Care Cancer 2000;8:472–478.
9. Detmar SB, Aaronson NK, Wever LD, Muller M, 28. Dexter PR, Wolinsky FD, Gramelspacher GP, et al: Ef-
Schornagel JH: How are you feeling? Who wants to fectiveness of computer-generated reminders for in-
know? Patients’ and oncologists’ preferences for dis- creasing discussions about advance directives and
cussing health-related quality-of-life issues. J Clin On- completion of advance directive forms. A randomized,
col 2000;18:3295–3301. controlled trial. Ann Intern Med 1998;128:102–110.
10. Dowsett SM, Saul JL, Butow PN, et al: Communica- 29. Roter DL, Larson S, Fischer GS, Arnold RM, Tulsky
tion styles in the cancer consultation: Preferences for JA: Experts practice what they preach: A descriptive
S-102 TULSKY

study of best and normative practices in end-of-life 46. Ferrell BR, Virani R, Grant M, et al: Evaluation of the
discussions. Arch Intern Med 2000;160:3477–3485. End-of-Life Nursing Education Consortium under-
30. Tulsky JA, Fischer GS, Rose MR, Arnold RM: Open- graduate faculty training program. J Palliat Med
ing the black box: How do physicians communicate 2005;8:107–114.
about advance directives? Ann Intern Med 1998; 47. Fallowfield L, Jenkins V, Farewell V, Saul J, Duffy A,
129:441–449. Eves R: Efficacy of a Cancer Research UK communi-
31. Hagerty RG, Butow PN, Ellis PA, et al: Cancer patient cation skills training model for oncologists: A ran-
preferences for communication of prognosis in the domised controlled trial. Lancet 2002;359:650–656.
metastatic setting. J Clin Oncol 2004;22:1721–1730. 48. Jenkins V, Fallowfield L: Can communication skills
32. Gattellari M, Voigt KJ, Butow PN, Tattersall MH: training alter physicians’ beliefs and behavior in clin-
When the treatment goal is not cure: Are cancer pa- ics? J Clin Oncol 2002;20:765–769.
tients equipped to make informed decisions? J Clin 49. Fallowfield L, Jenkins V, Farewell V, Solis-Trapala I:
Oncol 2002;20:503–513. Enduring impact of communication skills training:
33. Pfeifer MP, Mitchell CK, Chamberlain L: The value of Results of a 12–month follow-up. Br J Cancer
disease severity in predicting patient readiness to ad- 2003;89:1445–1449.
dress end-of-life issues. Arch Intern Med 2003;163: 50. Shilling V, Jenkins V, Fallowfield L: Factors affecting
609–612. patient and clinician satisfaction with the clinical con-
34. Blackhall LJ, Murphy ST, Frank G, Michel V, Azen S: sultation: can communication skills training for clin-
Ethnicity and attitudes toward patient autonomy. icians improve satisfaction? Psychooncology 2003;12:
JAMA 1995;274:820–825. 599–611.
35. Blackhall LJ, Frank G, Murphy ST, Michel V, Palmer 51. SUPPORT Principal Investigators: A controlled trial
JM, Azen SP: Ethnicity and attitudes towards life sus- to improve care for seriously ill hospitalized patients.
taining technology. Soc Sci Med 1999;48:1779–1789. The Study to Understand Prognoses and Preferences
36. Caralis PV, Davis B, Wright K, Marcial E: The influ- for Outcomes and Risks of Treatments (SUPPORT).
ence of ethnicity and race on attitudes toward ad- JAMA 1995;274:1591–1598.
vance directives, life-prolonging treatments, and eu- 52. Detmar SB, Muller MJ, Schornagel JH, Wever LD,
thanasia. J Clin Ethics 1993;4:155–165. Aaronson NK: Health-related quality-of-life assess-
37. Kagawa-Singer M, Blackhall LJ: Negotiating cross- ments and patient-physician communication: A ran-
cultural issues at the end of life: “You got to go where domized controlled trial. JAMA 2002;288:3027–3034.
he lives”. JAMA 2001;286:2993–3001. 53. Lilly CM, De Meo DL, Sonna LA, et al: An intensive
38. Finucane TE, Shumway JM, Powers RL, D’Alessandri communication intervention for the critically ill. Am
RM: Planning with elderly outpatients for contingen- J Med 2000;109:469–475.
cies of severe illness: A survey and clinical trial. J Gen 54. Lilly CM, Sonna LA, Haley KJ, Massaro AF: Intensive
Intern Med 1988;3:322–325. communication: Four-year follow-up from a clinical
39. Tierney WM, Dexter PR, Gramelspacher GP, Perkins practice study. Crit Care Med 2003;31:S394–S399.
AJ, Zhou XH, Wolinsky FD: The effect of discussions 55. Schneiderman LJ, Gilmer T, Teetzel HD, et al: Effect
about advance directives on patients’ satisfaction of ethics consultations on nonbeneficial life-sustain-
with primary care. J Gen Intern Med 2001;16:32–40. ing treatments in the intensive care setting: A ran-
40. Butow PN, Brown RF, Cogar S, Tattersall MH, Dunn domized controlled trial. JAMA 2003;290:1166–1172.
SM: Oncologists’ reactions to cancer patients’ verbal 56. Eide H, Frankel R, Haaversen AC, Vaupel KA, Grau-
cues. Psychooncology 2002;11:47–58. gaard PK, Finset A: Listening for feelings: identifying
41. McDonagh JR, Elliott TB, Engelberg RA, et al: Family and coding empathic and potential empathic oppor-
satisfaction with family conferences about end-of-life tunities in medical dialogues. Patient Educ Couns
care in the intensive care unit: Increased proportion 2004;54:291–297.
of family speech is associated with increased satis- 57. Caris-Verhallen W, Timmermans L, van Dulmen S:
faction. Crit Care Med 2004;32:1484–1488. Observation of nurse–patient interaction in oncology:
42. Levinson W, Roter D: The effects of two continuing Review of assessment instruments. Patient Educ
medical education programs on communication skills Couns 2004;54:307–320.
of practicing primary care physicians. J Gen Intern
Med 1993;8:318–324.
43. Smith RC, Lyles JS, Mettler J, et al: The effectiveness Address reprint requests to:
of intensive training for residents in interviewing. A James A. Tulsky, M.D.
randomized, controlled study. Ann Intern Med Center for Palliative Care
1998;128:118–126. Duke University
44. Sulmasy DP, Song KY, Marx ES, Mitchell JM: Strategies
Hock Plaza, Suite 1105
to promote the use of advance directives in a residency
outpatient practice. J Gen Intern Med 1996;11:657–663.
2424 Erwin Road
45. Maguire P, Pitceathly C: Improving the psychologi- Durham, NC 27705
cal care of cancer patients and their relatives. The role
of specialist nurses. J Psychosom Res 2003;55:469–474. E-mail: jtulsky@duke.edu