Detmar SB, Muller MJ, Wever LDV, Schornagel JH, Aaronson NK. Patient-Physician Communication During Outpatient Palliative Treatment VisitsAn Observational Study. JAMA. 2001;285(10):1351-1357. doi:10.1001/jama.285.10.1351
Author Affiliations: Division of Psychosocial Research and Epidemiology (Messrs Detmar, Muller, Wever, and Dr Aaronson) and Department of Medical Oncology (Dr Schornagel), The Netherlands Cancer Institute/Antoni van Leeuwenhoek Hospital, Amsterdam, the Netherlands.
The Patient-Physician Relationship Section Editor: Richard M. Glass, MD, Deputy Editor.
Context Improving health-related quality of life (HRQL) is an important goal
of palliative treatment, but little is known about actual patient-physician
communication regarding HRQL topics during palliative treatment.
Objectives To investigate the content of routine communication regarding 4 specific
HRQL issues between oncologists and their patients and to identify patient-,
physician-, and visit-specific factors significantly associated with discussion
of such issues.
Design Observational study conducted between June 1996 and January 1998.
Setting Outpatient palliative chemotherapy clinic of a cancer hospital in the
Participants Ten oncologists and 240 of their patients (72% female; mean age, 55
years) who had incurable cancer and were receiving outpatient palliative chemotherapy.
Main Outcome Measures Patient and physician questionnaires and audiotape analysis of communication
regarding daily activities, emotional functioning, pain, and fatigue during
an outpatient consultation using the Roter Interaction Analysis System.
Results Physicians devoted 64% of their conversation to medical/technical issues
and 23% to HRQL issues. Patients' communication behavior was divided more
equally between medical/technical issues (41%) and HRQL topics (48%). Of the
independent variables investigated, patients' self-reported HRQL was the most
powerful predictor of discussing HRQL issues. Nevertheless, in 20% to 54%
of the consultations in which patients were experiencing serious HRQL problems,
no time was devoted to discussion of those problems. In particular, these
patients' emotional functioning and fatigue were unaddressed 54% and 48% of
the time, respectively. Discussion of HRQL issues was not more frequent in
consultations in which tumor response was evaluated.
Conclusion Despite increasing recognition of the importance of maintaining patients'
HRQL as a goal of palliative treatment, the amount of patient-physician communication
devoted to such issues remains limited and appears to make only a modest contribution,
at least in an explicit sense, to the evaluation of treatment efficacy in
daily clinical practice.
The principal goals of palliative cancer treatment are to prolong the
patient's life and to alleviate debilitating symptoms of the disease.1- 3 In both cases, health-related
quality of life (HRQL) issues, defined in terms of the patient's physical,
emotional, and social functioning, and well-being4,5
may be of central importance in selecting among available treatment options
and in monitoring the effects of such treatments over time. An essential condition
for optimal palliative cancer treatment is that physicians communicate effectively
with their patients to obtain as complete a picture as possible of the patients'
physical and psychosocial health status.
Although, to the best of our knowledge, no studies have been directed
at patient-physician communication regarding HRQL in the palliative treatment
setting, there is a substantial amount of literature on the content of patient-physician
communication in other medical settings.6- 11
Results from these studies indicate that a number of elements play a role
in the way in which information is exchanged between physicians and their
patients. First, the nature of the symptoms themselves can influence patient-physician
communication. For example, Funch,10 investigating
the symptom-reporting behavior of colorectal cancer patients, found that 54%
of experienced symptoms were spontaneously reported, and that chronic, nonspecific,
and mild symptoms were less likely to be discussed than more acute, concrete,
and severe symptoms.
Second, patients vary in their willingness and ability to talk about
their problems and concerns. Female patients tend to ask more questions and
to receive more information than their male counterparts.12,13
Similar results have been reported for patients with relatively high levels
of education compared with those who are less well-educated.14
Physicians also vary widely in their interest in and their ability to
elicit relevant information from their patients.6,11,15
Levinson and Roter11 found that patients disclose
significantly more information about their emotional and social functioning
when their physician has a positive attitude toward the psychosocial aspects
of patient care. Maguire et al15 found that
patient disclosure of psychosocial information was inhibited when physicians
used closed-ended questions and when they focused their attention on physical
Finally, structural factors may play a role in the exchange of information.
For example, perceived time pressure may decrease the amount of information
Taken together, these studies indicate that patient-physician information
exchange may not always be optimal. In particular, nonspecific symptoms and
psychosocial issues are often left unaddressed.19- 21
Whether this also holds true for the palliative treatment setting is unclear.
Particularly in the realm of palliative treatment, there is increasing recognition
that HRQL is an outcome that is as important, if not more important, than
traditional outcomes such as survival.22,23
Thus, one might expect more discussion about HRQL issues in the palliative
treatment setting, and in particular during consultations in which the effect
of the treatment is being evaluated.
In a previous article we described the preferences of cancer patients
being treated with palliative chemotherapy and the attitudes and self-reported
behavior of their physicians toward discussing HRQL issues.24
In this article, we present results based on actual communication during outpatient
consultations. We chose to focus on 4 key HRQL issues affecting the broad
spectrum of patients receiving palliative treatment: daily activities, emotional
functioning, pain, and fatigue.25
The patient and physician samples were drawn from the outpatient clinics
of the Netherlands Cancer Institute. The physician sample consisted of the
physicians working in the department of medical oncology. The patient sample
comprised a consecutive series of patients with incurable cancer who were
receiving outpatient palliative chemotherapy and who were under the care of
one of the participating physicians. Inclusion criteria included the following:
older than 18 years, basic proficiency in Dutch, not participating in a concurrent
HRQL study, and having received at least 2 cycles of chemotherapy. This latter
criterion was applied to increase the likelihood that both the patients' HRQL
and, at least in some cases, evaluation of tumor response would be relevant
topics for discussion. The institutional review board of the hospital approved
Patients were invited to participate by means of a letter that explained
that the study was designed to obtain a better understanding of patient-physician
communication during the palliative treatment period. After providing informed
consent, patients and physicians were asked to complete several questionnaires.
The first subsequent outpatient consultation was audiotaped. The audiotaping
procedure was pilot-tested with all physicians and 2 of their patients. The
physicians and patients indicated that the audiotaping did not influence their
"natural" communication, which is in line with the literature.26
Patients who declined to participate were asked to respond to a brief questionnaire
Patients' Characteristics and Self-reported Health Status.
Patients' sociodemographic characteristics and their preferences for
discussing HRQL issues were obtained by means of a questionnaire.
Preferences were assessed in general terms, rather than in relation
to a specific consultation. The response categories were: "No, I would prefer
not to discuss this topic"; "Yes, but my physician has to initiate discussion
of this topic"; and "Yes, I want to discuss this topic."
The patients' self-reported HRQL was assessed by means of the COOP/WONCA
[Dartmouth Primary Care Cooperative Research Network/ World Organization of
National Colleges, Academies and Academic Associations of General Practitioners/Family
Physicians] functional health assessment charts assessing physical fitness,
emotional functioning, daily and social activities, overall health, and pain.27 Two additional charts assessing fatigue and overall
HRQL were also included. The time frame used was the previous 2-week period.
Response categories ranged from 1 (excellent) to 5 (very poor). Only the charts
assessing patients' daily activities, emotional functioning, pain, and fatigue
were used in the current analysis.
Physicians' Characteristics and Attitudes. Questionnaires were used to obtain data on physicians' sociodemographic
and professional characteristics and on their perceived responsibility for
discussing various HRQL issues with their patients.
Response categories for the latter questions included: "Discussing this
topic is completely my task"; "Discussing this topic is partially my task,
and partially the task of other health care providers"; and "Discussing this
topic is not my task."
Patient-Physician Communication. Analysis of the audiotaped consultations was conducted by means of an
adapted version of the Roter Interaction Analysis System, which places each
utterance of the patient and physician into a mutually exclusive category.28,29 The categories are organized into
3 conceptual clusters: process, affective, and content (Table 1).
The process cluster includes utterances intended to give direction to
the flow of the visit and facilitate communication. The affective cluster
consists primarily of statements that show involvement with the situation.
The content cluster includes all statements relating to medical/technical
issues, HRQL issues, and administrative issues.
Coding was carried out directly from audiotape by 3 trained raters.
All raters coded a random sample of 15% of the audiotapes to assess interrater
reliability. The mean interrater reliability was 0.87 (range, 0.66-0.99) for
physician categories, and 0.84 (range, 0.64-0.99) for patient categories.
These reliability estimates are comparable to those achieved in other studies.30- 32
Additionally, a content checklist was used to code whether the selected
HRQL topics were discussed at any time during the consultation, regardless
of the amount of time devoted to the topic and who had initiated discussion
of these topics.
Other Characteristics of the Consultations. The audiotapes were also used to determine if tumor response had been
discussed (labeled as an "evaluative" consultation).
Also noted were the duration of the consultation and any interruptions
that took place. Finally, the waiting time, an indirect indicator of the time
pressure under which the consultation took place, was calculated by subtracting
the scheduled visit time from the actual time at which the consultation took
Mean percentages of the total utterances devoted to the specific topics
of discussion were calculated. Although verbal utterances differed in length,
the correlation between the number of utterances and the consultation time
was high (r = 0.83). Therefore, the frequency of
utterances per topic corresponds approximately to the amount of time dedicated
to that topic. To control for differences in consultation time, percentages
were used instead of simple frequency counts.
Univariate statistics (t tests and χ2 statistics) were used to assess the association between patient, physician,
and consultation characteristics and communication about the selected HRQL
issues. For purpose of these analyses, patients' age was categorized into
2 groups (≤60 years and >60 years), the response categories of the COOP/WONCA
charts were collapsed into 2 categories (no to limited problems vs moderate
to severe problems), and the response categories of patients' preferences
and physicians' attitude were categorized into 2 groups (wanting to discuss,
either self- or physician-initiated, vs not wanting to discuss; full responsibility
vs partial or no responsibility, respectively).
Because patients were nested within physicians, we considered the use
of multilevel analysis in determining those factors related most strongly
to the discussion of the selected HRQL issues. However, multilevel analysis
requires a minimum of approximately 30 cases in the highest level.33 With only 10 physicians, the use of these methods
can lead to inconsistent parameter estimates. To examine differences between
physicians, intraclass correlations were computed for the selected outcome
measures. All intraclass correlations were nonsignificant and low (intraclass
correlation, <0.05), indicating that the variance components between physicians
were very small and thus that regression models for the total patient sample
would yield unbiased SEs. Therefore, both linear and logistic regression analyses
(forward, stepwise procedures) were used to test the simultaneous effect of
the independent variables. Variables were included in the regression models
if they were associated with the relevant outcomes at the univariate level
(P <.10). For all other statistical tests, the
significance level was set at P = .05.
All medical oncologists (n = 12) were invited to participate in the
study, and 10 agreed. The 2 oncologists who declined raised objections to
having their consultations audiotaped. Of the participating physicians, 4
were female, their mean age was 44 years with, on average, 11 years of work
experience in oncology (Table 2).
Between June 1996 and January 1998, a total of 382 patients were asked
to participate, of whom 273 (71%) agreed. Of the 109 nonrespondents, 50 declined
because they thought the study would be too burdensome, 43 indicated insufficient
interest or lack of time, and 16 expressed difficulty with the audiotaping.
A nonrespondent analysis indicated that patients who declined to participate
had less education (P<.001) and rated their overall
HRQL as significantly lower (mean = 3.7 vs 3.1, P<.05)
than the participants.
In 33 cases, the consultation was either not recorded due to logistical
reasons or the tape recording was of insufficient quality for analysis. Thus,
audiotaped consultations of 240 patients were available (between 19 and 30
per physician). Seventy-three percent of the patients were female, with a
mean age of 55 years. Patients had a wide range of cancer diagnoses, with
breast cancer being the most prevalent. The patients showed substantial variation
in scores on the COOP/WONCA charts. All patients had visited their physician
previously (mean = 9 previous visits) (Table 2).
The mean duration of the consultations was 16.09 minutes (range = 4.10-40.25,
SD = 6.82), with the physicians and patients exchanging a mean of 264 utterances
(range = 35-711, SD = 117). Overall, the physicians generated slightly more
discussion than the patients (53.5% vs 46.5%). Approximately one quarter of
the physicians' and patients' utterances were coded into the process category,
and approximately 13% of utterances conveyed emotional affect and social talk.
Slightly more than 60% of all utterances were substantive and fell into
the content category. The focus of interest is on the division of topics within
this content category. Approximately two thirds of the physicians' substantive
communication was medical/technical; for patients, this was 41%. Physicians
and patients devoted approximately one quarter and one half of their substantive
conversation, respectively, to HRQL issues, with symptoms (pain, fatigue,
and other symptoms combined) being most often discussed (Table 3).
As indicated in Table 4,
patients' daily functioning was discussed in approximately two thirds of the
consultations, most often initiated by the physician. Pain was discussed in
approximately three quarters of the consultations, with physicians taking
the lead half of the time. Fatigue and emotional problems were brought up
in 46% and 35% of the consultations, respectively, primarily by the patients.
The mean percentage of the physicians' substantive utterances devoted to these
4 topics was less than 5% per topic. For the patients, the range was from
2.1% (fatigue) to 10.1% (emotional functioning).
Emotional problems and pain were discussed significantly more often
during consultations with female than with male patients (38% vs 26% and 76%
vs 63%, respectively, P = .05). No other significant
associations were found between patients' sociodemographic characteristics
and HRQL communication.
Detailed results pertaining to patients' communication preferences are
described elsewhere.24 While the large majority
(between 80% and 95%) of the patients expressed a desire to discuss the various
HRQL topics, no significant associations were found between patients' preferences
and the frequency of actual discussion.
Table 5 shows the relationship
between patients' self-reported HRQL and HRQL communication. Emotional functioning,
pain, and fatigue were discussed significantly more frequently in those cases
where patients reported serious problems in these areas. Nevertheless, it
is noteworthy that among those patients experiencing serious emotional problems
or fatigue, these issues were discussed in only about half of the cases.
No statistically significant associations were observed between physicians'
age, sex, or years of work experience, and HRQL communication. However, the
physicians' attitude toward discussing their patients' emotional problems
was reflected in their actual communication behavior. More time was devoted
to discussing patients' emotional functioning by those physicians who felt
a strong responsibility to do so compared with those who did not (9% vs 6%
of total utterances, respectively, P = .04).
The effect of the treatment on tumor growth was addressed explicitly
in 60% of the consultations. No significant differences were found in HRQL
communication between consultations with or without such an evaluative component.
Forty percent of the patients were seen at their scheduled appointment
time. Patients' emotional problems were more likely to be discussed during
consultations held at the appointed time than during those that were delayed
(49% vs 32%, P = .05). In addition, more time was
devoted to this issue during the on-time consultations (9% vs 5.5% of utterances, P = .05). This could not be explained by differences in
the duration of on-time vs delayed consultations, which was approximately
the same (16.4 minutes vs 15.7 minutes, respectively). The length of the consultation,
interruptions, and the number of prior visits had no significant effect on
Multiple linear and logistic regression analyses were performed to test
the simultaneous effect of patient-, physician- and consultation-related characteristics
on the frequency and duration of communication about each of the 4 selected
HRQL issues. Patients' self-reported problems and consultation waiting time
emerged as significant predictors of the frequency with which patients' emotional
problems were discussed (χ25 = 15, P = .009). The percentage of time devoted to discussing emotional problems
was related significantly to patients' self-reported problems, consultation
waiting time, and the attitude of the physicians (F5 = 6.7, P<.001) The only variable found to be related significantly
at the multivariate level to the discussion of pain and fatigue was the severity
of these symptoms as reported by the patients. No variables emerged as significant
multivariate predictors of the discussion of patients' daily activities.
The objectives of this study were to describe the content of patient-physician
communication during outpatient palliative consultations, and to identify
factors that play an important role in the discussion of 4 key HRQL issues:
patients' daily activities, emotional functioning, pain, and fatigue.
In general, patients' communication behavior was directed as equally
to medical/technical issues (eg, results of blood tests, the dose of the chemotherapy)
as it was to HRQL topics. The content of physicians' communication was focused
primarily on biomedical/technical issues and, to a lesser degree, on HRQL
issues (a ratio of 3:1). This latter finding is in line with results found
in other medical settings,31,32
suggesting that there is not a substantial difference in HRQL communication
between palliative treatment settings and other treatment settings such as
primary care and curative treatment.
The discussion of HRQL issues varied depending on the specific topic
under consideration. Whereas patients' daily activities and pain were discussed
in the majority of consultations, fatigue was discussed in less than half,
and emotional functioning in only one third of the consultations. Importantly,
the same variation was observed for those consultations in which patients
experienced serious problems in one of these domains. Given that most patients
with cancer consider these issues to be of importance and wish to discuss
them with their physician,24,34- 36
these results suggest that patients' communication needs may often be left
Physicians' attitudes had an observable impact on HRQL communication.
Specifically, when physicians did not feel fully responsible for discussing
their patients' emotional functioning, it was significantly less likely to
be discussed. Additionally, patients' emotional problems were less frequently
discussed during consultations that took place behind schedule, probably reflecting
concern that their discussion would be too time-consuming. However, no significant
differences were found in the duration of consultations as a function of whether
or not emotional issues were discussed. Other studies have also failed to
demonstrate an effect of the discussion of psychosocial issues on the length
of medical consultations.37,38
Despite its high prevalence, fatigue was discussed relatively infrequently
and usually at the initiative of the patient. This finding is not unexpected
in that previous research has shown that symptoms of a nonspecific and chronic
nature are less likely to be discussed.10 Additionally,
until quite recently, few therapeutic options were available for combating
fatigue. Physicians may be reluctant to address what is often perceived to
be an intransigent problem.36
Consistent with the literature, we found that HRQL issues were discussed
significantly more often with female than with male patients. This likely
reflects sex differences in symptom awareness and reporting. In general, women
are more likely than men to rate themselves as being impaired and to report
symptoms.39 This is probably the reason that
in the regression analyses, after controlling for the severity of self-reported
problems, sex was no longer associated significantly with the discussion of
Contrary to expectations, the type of consultation (evaluative vs nonevaluative)
was not related significantly to HRQL communication. This suggests that HRQL
issues may play a lesser role, at least explicitly, in decisions surrounding
the (dis)continuation of treatment than one would expect given the importance
expressed about such issues by physicians.22,40,41
Efforts directed toward increasing the frequency with which HRQL issues are
discussed could enhance patient participation in decision-making, as patients
often withdraw from communication that is of a very narrow biomedical and
Several strategies have been proposed for increasing the likelihood
of HRQL issues being discussed during medical encounters. The effectiveness
of training programs to enhance physicians' communication skills and/or to
encourage patients to verbalize their needs and concerns has been demonstrated.43- 47
Recently, interest has been expressed in introducing formal HRQL assessments
into daily clinical practice as a means of triggering patient-physician communication.48,49 Promising results have been obtained
from several studies of such procedures, although they have been based on
Some caution should be exercised in interpreting the results of the
current study. First, the study was conducted in a single hospital and, although
the patient sample was large, the number of physicians involved was limited.
Second, all patients in the current study were white, reflecting the very
low incidence of cancer in the relatively young immigrant population in the
Netherlands. As ethnicity and culture can influence the patterns and content
of patient-physician communication, future studies are needed that use more
culturally diverse patient samples. Third, communication with regard to HRQL
issues was observed during a single visit to the outpatient clinic. It is
possible that patients and physicians had discussed these issues in one of
their earlier contacts and, therefore, may not have found it necessary to
discuss them again. However, one could argue that each medical encounter should
include at least a minimum amount of question-asking by the physician to track
changes in HRQL over time.
In conclusion, our results indicate that the likelihood that HRQL issues
will be addressed is greater for patients who experience relatively serious
problems, whose physician feels a sense of personal responsibility for discussing
such issues, and whose visit takes place at the scheduled time. In general,
however, only limited attention was paid to HRQL issues, in particular to
patients' emotional problems and fatigue, in these patient-oncologist interviews
during outpatient palliative treatment. Importantly, HRQL issues do not appear
to play an especially prominent role, at least not in an explicit sense, in
the evaluation of treatment efficacy.