Cancer patient preferences for quality and length of life
-
Upload
neal-j-meropol -
Category
Documents
-
view
214 -
download
2
Transcript of Cancer patient preferences for quality and length of life
Cancer Patient Preferences for Qualityand Length of Life
Neal J. Meropol, MD1
Brian L. Egleston, MPP, PhD2
Joanne S. Buzaglo, PhD2
Al B. Benson, III, MD3
Donald J. Cegala, PhD4,5
Michael A. Diefenbach, PhD6
Linda Fleisher, MPH2
Suzanne M. Miller, PhD2
Daniel P. Sulmasy, OFM, MD, PhD7
Kevin P. Weinfurt, PhD8
CONNECT Study Research Group
1 Divisions of Medical Science and PopulationScience, Fox Chase Cancer Center, Philadelphia,Pennsylvania.
2 Division of Population Science, Fox Chase Can-cer Center, Philadelphia, Pennsylvania.
3 Division of Hematology/Oncology, Robert H.Lurie Comprehensive Cancer Center, Northwes-tern University, Chicago, Illinois.
4 Department of Communication, Ohio State Uni-versity, Columbus, Ohio.
5 Department of Family Medicine, College ofMedicine, Ohio State University, Columbus, Ohio.
6 Department of Urology, Mount Sinai MedicalCenter, New York, New York.
7 Bioethics Institute, New York Medical College,Valhalla, New York.
8 Department of Psychology and Neuroscience,Duke University School of Medicine, Duke Univer-sity, Durham, North Carolina.
BACKGROUND. Optimal patient decision making requires integration of patient
values, goals, and preferences with information received from the physician. In
the case of a life-threatening illness such as cancer, the weights placed on quality
of life (QOL) and length of life (LOL) represent critical values. The objective of
the current study was to describe cancer patient values regarding QOL and LOL
and explore associations with communication preferences.
METHODS. Patients with advanced cancer completed a computer-based survey
before the initial consultation with a medical oncologist. Assessments included
sociodemographics, physical and mental health state, values regarding quality
and length of life, communication preferences, and cancer-related distress.
RESULTS. Among 459 patients with advanced cancer, 55% placed equal valued on
QOL and LOL, 27% preferred QOL, and 18% preferred LOL. Patients with a QOL
preference had lower levels of cancer-related distress (P < .001). A QOL preference
was also associated with older age (P 5 .001), male sex (P 5 .003), and higher edu-
cational level (P 5 .062). Patients who preferred LOL over QOL desired a more sup-
portive and less pessimistic communication style from their oncologists.
CONCLUSIONS. These data indicate that a values preference for LOL versus QOL
may be simply measured, and is associated with wishes regarding the nature of
oncologist communication. Awareness of these values during the clinical encoun-
ter could improve decision making by influencing the style and content of
the communication between oncologists and their patients. Cancer 2008;113:
3459–66. � 2008 American Cancer Society.
KEYWORDS: quality of life, cancer communication, physician-patient communication,patient preferences, communication preferences, cancer-related distress, length of lifepreferences, patient decision making, cancer communication aid, patient values.
Supported by R01CA082085 and Fox Chase Can-cer Center Population Studies Facility and Behav-ioral Research Core Facility (P30CA06927) fromthe National Cancer Institute.
Clinicaltrials.gov registration ID: NCT00244868.Available at: http://www.clinicaltrials.gov/ct/show/NCT00244868?order56
We thank the following clinical sites and indivi-duals for enrolling patients in this study: FoxChase Cancer Center (Principal Investigator: Neal
J. Meropol, MD; Research Assistants: NicholasSolarino and Jonathan Trinastic); Robert H. LurieComprehensive Cancer Center, NorthwesternUniversity (Principal Investigator: Al B. BensonIII, MD; Research Assistants: Sheano Gold andLisa Stucky-Marshall); and Meharry Medical Col-lege (Principal Investigator: Steven N. Wolff, MD;Research Assistants: Kim Burlison and RoslynHatchett-Pope).
The CONNECT Study Research Group includes:Andrew Balshem, BA (Fox Chase Cancer Center,Philadelphia, Penn); Ellyn Micco, MS (Fox ChaseCancer Center, Philadelphia, Penn); Jennifer L.Millard, BS (Fox Chase Cancer Center, Philadel-phia, Penn); Eric A. Ross, PhD (Fox Chase Can-cer Center, Philadelphia, Penn); Kevin A.
Schulman, MD (Duke University School of Medi-cine, Durham, NC); Elyse Slater, MS (Fox ChaseCancer Center, Philadelphia, Penn); NicholasSolarino, MS (Fox Chase Cancer Center, Phila-delphia, Penn); and Jonathan Trinastic, BS (FoxChase Cancer Center, Philadelphia, Penn), all ofwhom contributed to the authorship of this arti-cle.
Address for reprints: Neal J. Meropol, MD,Department of Medical Oncology, Fox ChaseCancer Center, 333 Cottman Avenue, Philadel-phia, PA 19111; Fax: (215) 728-3639; E-mail:[email protected]
Received April 11, 2008; revision received June11, 2008; accepted June 30, 2008.
ª 2008 American Cancer SocietyDOI 10.1002/cncr.23968Published online 5 November 2008 in Wiley InterScience (www.interscience.wiley.com).
3459
P atients with advanced cancer exist in a unique
medical context in which they are facing mortal-
ity and may be considering treatment options that
have significant potential for toxicity. In addition,
therapeutic choices are characterized by uncertain
outcomes, and may be varied and complex, includ-
ing supportive care alone, standard treatments (eg,
chemotherapy, radiotherapy, and biologic therapy),
and investigational approaches. Quality patient deci-
sion making requires an adequate patient under-
standing of treatment options, including potential
benefit and harm. The physician serves as the pri-
mary source of medical information for cancer
patients1,2; as such, the communication between
physician and patient is of critical importance to
quality decision making.3-5
In a survey of cancer patients who were consid-
ering participation in phase 1 trials, we observed dis-
cordance between patient and oncologist perceptions
regarding the content of their consultations.6 In par-
ticular, patients were far less likely to report discus-
sion of quality of life (QOL) issues than their
physicians. This lack of agreement is a potential
source of concern insofar as nearly all of the patients
taking part in the survey valued QOL at least as
highly as length of life (LOL).
Adequate communication regarding the impact
of treatment on QOL is of particular importance
given that patient preference for either QOL or LOL
can influence patient treatment decision making. For
example, among cancer patients with advanced dis-
ease, an individual’s preference for LOL over QOL is
associated with treatment preference for chemother-
apy over watchful waiting.7 Furthermore, several
sociodemographic factors are reported to be asso-
ciated with a preference for QOL or LOL. A prefer-
ence for QOL is associated with older age8,9 and
having no children.7,8 In contrast, a preference for
LOL is associated with being young, having children,
and good functional health status.8,10 Despite its im-
portance for cancer patient decision making regard-
ing treatment, to our knowledge few studies to date
have explored how an individual’s preference for
QOL or LOL influences the way in which patients
wish their physicians to present prognostic and treat-
ment-related information.
To approach the understanding of how patient
preferences impact decision making, we applied the
Cognitive-Social Health Information Processing (C-
SHIP) model.11,12 This is a comprehensive framework
that delineates the cognitive and affective factors
involved in health information processing.11,12 These
factors include the individual’s self-construals (eg,
perceived vulnerability to disease), affect (eg, anxious
preoccupation with cancer), and values and goals
(eg, concerning the physician-patient encounter and
the importance of QOL), as well as goals related to
achieving desired treatment outcomes (eg, expressing
one’s concerns and values).11 The C-SHIP model
emphasizes the role of distress in information proc-
essing and decision making.13 Because distress can
impact a patient’s ability to process critical prognos-
tic and treatment-related information relevant to
treatment choice,14 examination of its relation to an
individual’s values (eg, importance of QOL and LOL)
and communication preferences is essential.
In an effort to improve the matching of consulta-
tion content with individual patient values, we devel-
oped a computer-based communication aid. We
currently are conducting a randomized clinical trial
to assess the impact of this intervention on the con-
tent of oncology consultations and patient satisfac-
tion with physician communication. In this study, we
describe the baseline characteristics of the study
population in an effort to define the cognitive-affec-
tive profiles and communication preferences among
those cancer patients who prefer QOL and those who
prefer LOL.
MATERIALS AND METHODSPatient SelectionThis report describes baseline survey data obtained
as part of an intervention study to evaluate the effi-
cacy of a computer-based communication aid on
treatment decision making in patients with advanced
cancer. The parent study is a 3-arm randomized clin-
ical trial. Two of the arms include the survey items
that form the basis for this report.
Participants were patients with advanced cancer
being treated at 3 academic cancer centers in the
US. Eligibility criteria included: 1) first outpatient
consultation with a medical oncologist; 2) documen-
ted metastatic malignancy; 3) aged �18 years; 4)
ability to read and verbally communicate in English;
and 5) ability to provide written informed consent to
participate. Potential study participants were ascer-
tained between January 2005 and January 2007
through a review of new patient schedules and medi-
cal information forwarded at the time of scheduling,
before the initial medical oncologist consultation.
ProcedureAfter ascertainment, consent to contact potential
participants was obtained from the physician with
whom the patient was scheduled. A research assist-
ant then contacted patients by telephone to explain
the study. After providing verbal consent, participants
3460 CANCER December 15, 2008 / Volume 113 / Number 12
were given the option to complete the computer-
based survey at home, using a secure web-inter-
face,15 or arrive 1 hour before their appointment to
complete the survey at the clinical site. Participants
using the secure website to complete the survey
signed informed consent documents electronically.
Participants were given a toll-free number for techni-
cal support and to ask any questions regarding the
study. All participants provided written consent at
the time of their arrival for their physician appoint-
ments. The study was approved by the Institutional
Review Boards at the participating centers.
MeasuresPatient characteristics included sex, age, marital sta-
tus, education, employment status, race, ethnicity,
cancer diagnosis, and treatment history.
Physical and emotional health state was assessed
using the Short-Form (SF)-12 and the Revised Impact
of Events Scale (RIES). The SF-1216 is a highly reli-
able shortened version of the SF-3617 and is intended
to provide indices of QOL similar to the longer
instrument while reducing participant burden. Sub-
scales include a Physical Component Scale (PCS)
and Mental Component Scale (MCS). Cancer-related
distress was measured using the RIES, a well-vali-
dated 15-item questionnaire measuring event-related
intrusion and avoidance.18
QOL and LOL preferences were assessed with 3
items to determine the relative value that an indivi-
dual assigns to QOL and LOL. This instrument,
designed and refined based on prior research6,19 with
the target population, asked participants to select
from among 4 choices regarding whether QOL or
LOL was more important (QOL is all that matters,
QOL is more important but LOL matters, LOL is
more important but QOL matters, LOL is all that
matters). Participants were also asked to rate the im-
portance of QOL and LOL as independent domains
on 5-point scales (not at all, somewhat, moderately,
quite a bit, very important).
Communication preferences were assessed using
an 8-item survey developed based on a review of rel-
evant literature and existing physician-communica-
tion assessments.20-24 Participants were asked to rank
their agreement with each of these 8 preferences
regarding physician communication about prognos-
tic and probabilistic information on a 5-point scale
from ‘‘Strongly Agree’’ to ‘‘Strongly Disagree.’’
Statistical ConsiderationsWe defined QOL and LOL preferences in 2 ways. We
initially defined QOL versus LOL preference based
on the single 4-point survey item that required
patients to prioritize QOL and LOL. In an effort to
discriminate patient preferences further given infre-
quent selection of the extreme values of this single-
item measure, we defined 3 preference groups based
on independent rating of QOL and LOL on the 2 5-
point scales. For this composite analysis, we categor-
ized patients with regard to whether they selected
equal ratings for QOL and LOL on the independent
measures, or whether they rated 1 of these prefer-
ences higher on 1 of the measures. The analyses
described below use the composite measure.
For unadjusted analyses comparing the QOL
preference groups with potential confounders, we
used multiple linear regressions with robust standard
errors of the continuous confounders to assess statis-
tically significant differences among groups. We used
robust standard errors to weaken the analysis of var-
iance (ANOVA) assumption of normally distributed
errors. We used Wald tests of the regression parame-
ters to test the null hypothesis that there were no dif-
ferences in mean responses among the 3 QOL
preference groups. For categoric confounders, we
used the Fisher exact test. We used correlations to
assess the relation between the communication pref-
erence variables and the RIES distress score.
To adjust for age, education, sex, race, SF-12
PCS, and MCS scores in analyses relating QOL pre-
ferences with communication preferences and dis-
tress, we used propensity score adjustment through
propensity score-based weighting with doubly robust
estimation.25 We calculated robust standard errors
using the sandwich estimator of Huber26; our stand-
ard errors account for the uncertainty associated
with the estimation of the propensity scores. Propen-
sity score-based weighting is an effective means of
adjusting for potential confounders because it allows
for the creation of an adjusted population in which
the balance of confounders between QOL preference
groups can be assessed directly. If the average covari-
ate characteristics between QOL preference groups
in the adjusted population are similar, then it is unli-
kely that the covariates are confounding the adjusted
inferences.
The adjusted mean communication preference
scores presented in Table 1 and adjusted mean dis-
tress scores presented in Figure 1 represent estimates
of mean responses in the adjusted population.
To develop the propensity score model, we used
a multinomial logistic regression of the nominal 3-
category QOL variable. The age, SF-12, education,
sex, and race terms were entered into the model
using restricted cubic splines for age, SF-12 PCS, and
SF-12 MCS27 and interactions of sex with age and
education. The appropriateness of the model was
Cancer Patient Preferences/Meropol et al 3461
measured by it ability to balance the covariates
among the 3 groups.
RESULTSParticipant CharacteristicsA total of 1932 patients were contacted and 1101
(57%) agreed to be assigned a password to consider
participation in this 3-arm study; 743 patients (68%)provided informed consent and completed baselinesurveys. In all, 471 patients were randomized tothe 2 intervention arms, and 12 patients with miss-ing data were not included in these analyses.Therefore, 459 patients form the basis of the cur-rent study. Although all 459 patients were used toderive propensity scores, each communicationpreference response had an additional 1 or 2 miss-ing values. Patient demographics are shown inTable 2. The sample was primarily white (91%),51% male, and well-educated (70% with somecollege or more).
Preferences regarding QOL and LOL are summar-
ized in Table 3. Approximately half of the patient
participants (55%) equally valued QOL and LOL
based on the composite measure. Of those patients
with a preference, QOL was selected more often
than LOL (27% vs 18%). This finding is consistent
with the selection of QOL as more important on the
single-item scale (80% of patients) in which patients
were required to commit to a preference. Older age
(P < .001) and male sex (P < .003) were found to
be positively associated with a preference for QOL
(Table 4). To remove possible confounding by these
variables in the subsequent communication prefer-
ence analyses, we used propensity score adjustment.
In the propensity-weighted population, older age and
male sex were no longer associated with a preference
for QOL, indicating successful adjustment. In the
adjusted population, all P values were >.20. Com-
TABLE 1Relation Between Communication Preferences, Distress, and LOL/QOL Preference*
Total RIES Adjusted Mean (SE) P
Correlation P QOL Equal LOL P
I want the doctor to speak in a positive manner 0.096 3.65 4.03 4.19 <.001
.040 (.09) (.05) (.09)
I want to hear general terms (for example, ‘‘the treatment is likely to work’’) rather than
statistics (for example, ‘‘the treatment has a 75% likelihood of working’’).
0.034 3.22 3.63 3.76 .002
.468 (.12) (.07) (.12)
I want the doctor to soften the blow when giving me bad news 0.321 2.17 2.74 2.93 <.001
<.001 (.10) (.08) (.16)
I want the doctor to speak to me in an emotionally supportive way 0.289 3.83 4.15 4.24 <.001
<.001 (.08) (.04) (.07)
I want to hear detailed statistics 0.074 3.87 3.96 3.81 .358
.112 (.10) (.06) (.10)
I want the doctor to speak matter-of-factly (for example, give me the cold hard facts) 20.217 3.92 3.82 3.88 .721
<.001 (.10) (.07) (.10)
I want to hear averages about people like me 0.034 4.05 4.01 4.01 .910
.463 (.07) (.05) (.10)
I want to hear the doctor’s opinion about my case in particular 20.087 4.70 4.66 4.64 .788
.064 (.05) (.04) (.07)
LOL indicates length of life preferred; QOL, quality of life preferred; RIES, Revised Impact of Events Scale; SE, standard error.
*Scores represent average responses. A score of 1 indicates Strongly Disagree and a score of 5 indicates Strongly Agree (standard deviation).
FIGURE 1. Cancer-related distress and quality of life (QOL) versus lengthof life (LOL) preferences. The center bars represent the means of the pro-
pensity score-adjusted total Revised Impact of Events Scale (RIES) measure-
ments, whereas the whiskers represent the 95% confidence intervals.
P < .001 for a hypothesis test of the equality of the 3 means.
3462 CANCER December 15, 2008 / Volume 113 / Number 12
pared with patients with other diseases, those with
prostate cancer preferred QOL (P 5 .021) (Fig. 2).
Given a potential relation between affect and
values, an association between patient distress and
QOL versus LOL preferences was investigated. As
shown in Figure 1, those patients with a preference
for QOL had lower levels of cancer-related distress
than those patients with an LOL preference or no
preference. There was a marginally significant asso-
ciation noted between the mental health component
summary score of the SF–12 (P 5 .069) and the phys-
ical health component summary score of the SF–12
(P 5 .073), and preferences for QOL or LOL, with in-
ferior QOL on these measures associated with a pref-
erence for LOL (Table 4). There was no association
noted between length of time since diagnosis or pre-
vious receipt of systemic therapy, and QOL versus
LOL preferences.
We next explored whether QOL/LOL preferences
are associated with preferences for communication
with the physician. As shown in Table 1, after propen-
sity score-weighted adjustment, analyses indicate sig-
nificant differences in preferences for the style of
communication between those patients preferring
QOL and those preferring LOL. Specifically, compared
with the other groups, individuals who indicated that
they prefer LOL over QOL (testing for the equality of 3
means) were more likely to indicate that they would
like the physician to speak more positively (P < .001), to
want the physician to use general terms (P < .002), to
want the physician to soften bad news (P < .001), and
to want the physician to speak in an emotionally sup-
portive way (P < .001).
TABLE 2Patient Demographics (N5459)
No. %
Median age (range), y 60 (26-89)
Male 232 51%
Race
White 416 91%
Black 29 6%
Other 14 3%
Ethnicity
Non-Hispanic 452 99%
Hispanic 6 1%
Education
High school 136 30%
Some college 145 32%
�College degree 178 39%
Cancer diagnosis
Lung 83 18%
Colorectal 68 15%
Breast 53 12%
Pancreas 32 7%
Prostate 33 7%
Renal 24 5%
Ovarian 21 5%
Other* 145 32%
Y since diagnosis
Median 1
Mean (SD) 3.1 (4.3)
Range 0-41
Prior systemic therapy 276 60%
SD indicates standard deviation.
*‘‘Other’’ includes cancers accounting for <5% of all diagnoses within the study population, includ-
ing bladder, bone, cervical, endometrial, esophageal, head and neck cancers, lymphoma, melanoma,
stomach, and thyroid.
TABLE 3Preferences for Quality and Length of Life
Single-Item Measure No. %
Quality of life is all that matters 67 15%
Quality of life is more important, but length of life matters 299 65%
Length of life is more important, but quality of life matters 89 19%
Length of life is all that matters 4 1%
Composite Measure Mean SD
How important is quality of life to you? 4.4 0.8
How important is length of life to you? 4.2 1.0
No. %
Quality of life more important 123 27%
Equally important 252 55%
Length of life more important 84 18%
SD indicates standard deviation.
TABLE 4Relation Between Patient Characteristics and Values Preferences
Values Preference
QOL Equal LOL P
No. 123 252 84
Mean age (SD), y 62.6 (10.5) 59.2 (11.9) 57.3 (11.9) .001
PCS-12 43.6 (10.6) 44.7 (11.1) 41.7 (10.4) .073
MCS-12 46.1 (6.8) 44.6 (7.5) 43.9 (9.2) .069
Male 34% 49% 17% .003
Female 20% 61% 19%
Race .67
Black 17% 59% 24%
White 27% 55% 18%
Other 35% 50% 14%
Education .062
�High school 18% 62% 19%
Some college 28% 51% 21%
�College 32% 52% 16%
QOL indicates quality of life preferred; LOL, length of life preferred; SD, standard deviation; PCS,
Physical Component Scale; MCS, Mental Component Scale.
Cancer Patient Preferences/Meropol et al 3463
DISCUSSIONThere is great variability reported in cancer patients’
preferences regarding the content and format of
communication from their physicians.28-31 Matching
communication to patient preferences contributes to
quality patient decision making and satisfaction.23,32
Thus, tools to assist physicians in identifying relevant
patient preferences and guiding communication
accordingly could improve clinical outcomes. The
data we present indicate that a values preference for
LOL versus QOL may be simply measured, and is
associated with a desire for more supportive and less
pessimistic communication from the oncologist.
Communication skill in the cancer context is par-
ticularly critical given that patients are commonly
facing mortality and ‘bad news,’ treatment outcomes
are characterized by uncertainty, and treatment is
associated with the significant potential for morbidity.
Previous reports have identified a variety of patient
characteristics that bear on their wishes regarding
physician communication. For example, women and
patients with higher levels of educational attainment
have been shown to want more detailed information
regarding their prognosis.31,33,34 Female sex is also
associated with a desire for a supportive communica-
tion style over a blunter approach, whereas patients
with more education31 and older patients35 have
been shown to prefer a more fact-oriented style of
communication. The data we present support the hy-
pothesis that a preference for QOL versus LOL is a
key value that impacts treatment goals and desires
regarding physician communication. We also ob-
served that older age (P 5 .001), male sex (P 5 .003),
and higher educational level (P 5 .063) were asso-
ciated with a preference for QOL. Even after propen-
sity score adjustment, the QOL/LOL preference was
predictive of patient communication preferences.
It is notable that patients with less cancer-related
distress were more likely to favor QOL over LOL. The
direction of causation in this relationship cannot be
concluded from these data. It is possible that
increased distress is associated with greater difficulty
in processing QOL issues when faced with a life-
threatening illness, and therefore a focus on LOL is
preferred. It is also plausible that a greater concern
for one’s LOL leads to greater anxiety in the context
of an immediate threat to longevity. In either case,
high levels of distress can negatively impact risk in-
formation processing and communication, and, ulti-
mately, decision making. The finding that those
patients who favor LOL seek a more supportive com-
munication style and content is consistent with a
desire to minimize additional distress in this patient
subgroup. Furthermore, an alternative explanation
exists: patients who rely more heavily on blunting or
denial as a way to manage their distress may be
more inclined to avoid thinking about the end of life
and report a preference for LOL without fully consid-
ering the range of options, namely, shorter but
enhanced QOL. This subgroup of individuals may be
more inclined to prefer provider communications
that ‘soften the blow’ and avoid discussion of a diffi-
cult reality, and experience heightened distress when
provided with too much threat-relevant informa-
tion.36 Further research is required to more finely
delineate the relation between these cognitive and
affective influences on downstream events.
Although patients who value LOL over QOL pre-
fer a more supportive communication style, this does
not suggest that communication of ‘bad news’ should
be avoided. The withholding of negative prognostic
information may result in a loss of trust in the physi-
cian, decreased compliance, communication barriers
between partners, patient isolation and loss of con-
trol, and a lost opportunity to adapt to new circum-
stances.37-39 Furthermore, recent data suggest that
prognostic disclosure may be associated with
increased hope, even in the face of a poor progno-
sis.40 The data from the current study support the
recommendation that communication must be tai-
lored such that medical information is conveyed in a
manner that serves the cognitive and affective needs
of individual patients.28,37-39
The current study findings also demonstrate that
patients who value LOL over QOL have lower scores
on indicators of mental health, with a trend for these
patients to score lower on indicators of physical
health. We previously reported that the decision to
FIGURE 2. Quality of life (QOL)-length of life (LOL) preferences amongpatients with different types of cancer.
3464 CANCER December 15, 2008 / Volume 113 / Number 12
take part in an experimental therapy program (per-
haps indicative of an LOL focus) was related to a
patient’s reference point regarding the threat that
their cancer posed to their quality–adjusted sur-
vival.41 Further exploration of the relation between
health status and preferences regarding QOL and
LOL is warranted.
Because the study population in this report is
racially homogenous and relatively well–educated, it
is difficult to generalize these results to a more
diverse population. It is possible that the relation-
ships we observed between distress, QOL and LOL
preferences, and communication preferences may
differ in various socioeconomic or racial groups.
Furthermore, this study is limited in that the assess-
ment was conducted at a single timepoint, and does
not capture potential variations in values and goals
over the course of illness. Finally, the method we
used to classify patients’ preferences for LOL versus
QOL assumed that a 1-point difference on a 5-point
scale represents a precise, meaningful difference. If
such differences were not meaningful, there would
be greater noise in our classification of patients’ pref-
erence, thus leading to an underestimation of the
relationships reported here. It is important to note
that before this classification method is adopted as a
component of routine care, further work to establish
its discriminant ability is appropriate. Nevertheless,
to our knowledge this study represents 1 of the lar-
gest surveys of cancer patient preferences reported
to date, and patient ascertainment procedures sought
to identify and enroll all eligible subjects.
In summary, the results of the current study pro-
vide evidence that patient differences in the value
placed on QOL and LOL can be easily assessed, and
these values are associated with preferences for the
content and format of physician communication. We
currently are conducting a prospective randomized
clinical trial (NCT00244868) to assess whether pro-
viding patients with communication skills training,
and reporting patient values and preferences to the
oncologist before the initial consultation will impact
communication content and format, and improve
patient satisfaction with the consultation.
REFERENCES1. Engelman KK, Perpich DL, Peterson SL, Hall MA, Ellerbeck
EF, Stanton AL. Cancer information needs in rural areas.
J Health Commun. 2005;10:199-208.
2. Liang W, Burnett CB, Rowland JH, et al. Communication
between physicians and older women with localized breast
cancer: implications for treatment and patient satisfaction.
J Clin Oncol. 2002;20:1008-1016.
3. Sepucha KR, Belkora JK, Tripathy D, Esserman LJ. Building
bridges between physicians and patients: results of a pilot
study examining new tools for collaborative decision mak-
ing in breast cancer. J Clin Oncol. 2000;18:1230-1238.
4. Fallowfield L, Jenkins V. Effective communication skills arethe key to good cancer care. Eur J Cancer. 1999;35:1592-1597.
5. Fallowfield L, Jenkins V, Farewell V, Solis-Trapala I. Endur-ing impact of communication skills training: results of a12-month follow-up. Br J Cancer. 2003;89:1445-1449.
6. Meropol NJ, Weinfurt KP, Burnett CB, et al. Perceptions of
patients and physicians regarding phase I cancer clinical
trials: implications for physician-patient communication.
J Clin Oncol. 2003;21:2589-2596.
7. Koedoot CG, de Haan RJ, Stiggelbout AM, et al. Palliative
chemotherapy or best supportive care? A prospective study
explaining patients’ treatment preference and choice. Br J
Cancer. 2003;89:2219-2226.
8. Stiggelbout AM, de Haes JC, Kiebert GM, Kievit J, Leer JW.
Tradeoffs between quality and quantity of life: develop-
ment of the Questionnaire for Cancer Patient Attitudes.
Med Decis Making. 1996;16:184-192.
9. Voogt E, van der Heide A, Rietjens JA, et al. Attitudes of
patients with incurable cancer toward medical treatment
in the last phase of life. J Clin Oncol. 2005;23:2012-2019.
10. Rietjens JA, van der Heide A, Voogt E, Onwuteaka-Philipsen
BD, van der Maas PJ, van der Wal G. Striving for quality or
length at the end-of-life: attitudes of the Dutch general
public. Patient Educ Couns. 2005;59:158-163.
11. Miller SM, Shoda Y, Hurley K. Applying cognitive-social
theory to health-protective behavior: breast self-examina-
tion in cancer screening. Psychol Bull. 1996;119:70-94.
12. Miller SM, Diefenbach MA. The Cognitive-Social Health In-
formation Processing (C-SHIP) model: a theoretical frame-
work for research in behavioral oncology. In: Krantz D, ed.
Perspectives in Behavioral Medicine. Mahwah, NJ: Law-
rence Erlbaum; 1998:219-244.
13. Diefenbach MA, Miller SM, Daly MB. Specific worry about
breast cancer predicts mammography use in women at
risk for breast and ovarian cancer. Health Psychol. 1999;18:
532-536.
14. Erblich J, Montgomery GH, Valdimarsdottir HB. Biased
cognitive processing of cancer-related information among
women with family histories of breast cancer: evidence
from a cancer stroop task. Health Psychol. 2003;22:235-244.
15. Buzaglo JS, Millard JL, Ridgway CG, et al. An internet
method to assess cancer patient information needs and
enhance doctor-patient communication: a pilot study.
J Cancer Educ. 2007;22:233-240.
16. Ware J Jr, Kosinski M, Keller SD. A 12-item short-form
health survey: construction of scales and preliminary tests
of reliability and validity. Med Care. 1996;34:220-233.
17. Ware JE, Snow KK, Kosinski M, Gandek B. SF-36 Health
Survey: Manual and Interpretation Guide. Boston: Nimrod
Press; 1993.
18. Horowitz M, Wilner N, Alvarez W. Impact of event scale: a
measure of subjective stress. Psychosom Med. 1979;41:209-
218.
19. Weinfurt KP, Castel LD, Li Y, et al. The correlation between
patient characteristics and expectations of benefit from
phase I clinical trials. Cancer. 2003;98:166-175.
20. Bilodeau BA, Degner LF. Information needs, sources of in-
formation, and decisional roles in women with breast can-
cer. Oncol Nurs Forum. 1996;23:691-696.
21. Hack TF, Degner LF, Dyck G. Relationship between prefer-
ences for decisional control and illness information among
women with breast cancer. Soc Sci Med. 1994;39:279-289.
Cancer Patient Preferences/Meropol et al 3465
22. Detmar SB, Aaronson NK, Wever LDV, Muller M, Schorna-
gel JH. How are you feeling? Who wants to know?
Patients’ and oncologists’ preferences for discussing
health-related quality-of-life issues. J Clin Oncol. 2000;18:
3295-3301.
23. Dowsett SM, Saul JL, Butow PN, Dunn SM, Boyer MJ, Fin-
dlow R. Communication styles in the cancer consultation:
preferences for a patient-centered approach. Psychooncol-
ogy. 2000;9:147-156.
24. Leighl NB, Gattellari M, Butow PN, Brown R, Tattersall
MHN. Discussing adjuvant cancer therapy. J Clin Oncol.
2001;19:1768-1778.
25. Lunceford JK, Davidian M. Stratification and weighting
via the propensity score in estimation of causal treatment
effects: a comparative study. Stat Med. 2004;23:2937-2960.
26. Huber PJ. Robust estimation of a location parameter. Ann
Math Stat. 1964;35:73-101.
27. Harrell F. Regression Modeling Strategies. New York:
Springer; 2001.
28. Helft PR. Necessary collusion: prognostic communication
with advanced cancer patients. J Clin Oncol. 2005;23:3146-
3150.
29. Elkin EB, Kim SH, Casper ES, Kissane DW, Schrag D. Desire
for information and involvement in treatment decisions:
elderly cancer patients’ preferences and their physicians’
perceptions. J Clin Oncol. 2007;25:5275-5280.
30. Kutner JS, Steiner JF, Corbett KK, Jahnigen DW, Barton PL.
Information needs in terminal illness. Soc Sci Med. 1999;48:
1341-1352.
31. Butow PN, Maclean M, Dunn SM, Tattersall MH, Boyer MJ.
The dynamics of change: cancer patients’ preferences for
information, involvement and support. Ann Oncol. 1997;8:
857-863.
32. Schofield PE, Butow PN, Thompson JF, Tattersall MH,
Beeney LJ, Dunn SM. Psychological responses of patients
receiving a diagnosis of cancer. Ann Oncol. 2003;14:48-
56.
33. Parker PA, Baile WF, de Moor C, Lenzi R, Kudelka AP,
Cohen L. Breaking bad news about cancer: patients’ prefer-
ences for communication. J Clin Oncol. 2001;19:2049-2056.
34. Butow PN, Kazemi JN, Beeney LJ, Griffin AM, Dunn SM,
Tattersall MH. When the diagnosis is cancer: patient com-
munication experiences and preferences. Cancer. 1996;77:
2630-2637.
35. Fujimori M, Akechi T, Morita T, et al. Preferences of cancer
patients regarding the disclosure of bad news. Psychooncol-
ogy. 2007;16:573-581.
36. Miller SM. Monitoring versus blunting styles of coping
with cancer influence the information patients want and
need about their disease. Implications for cancer screening
and management. Cancer. 1995;76:167-177.
37. Girgis A, Sanson-Fisher RW. Breaking bad news: consensus
guidelines for medical practitioners. J Clin Oncol. 1995;13:
2449-2456.
38. Fallowfield LJ, Jenkins VA, Beveridge HA. Truth may hurt
but deceit hurts more: communication in palliative care.
Palliat Med. 2002;16:297-303.
39. Baile WF, Beale EA. Giving bad news to cancer patients:
matching process and content. J Clin Oncol. 2003;21(9
suppl):49s-51s.
40. Mack JW, Wolfe J, Cook EF, Grier HE, Cleary PD, Weeks JC.
Hope and prognostic disclosure. J Clin Oncol. 2007;25:
5636-5642.
41. Gaskin DJ, Weinfurt KP, Castel LD, et al. An exploration of
relative health stock in advanced cancer patients. Med
Decis Making. 2004;24:614-624.
3466 CANCER December 15, 2008 / Volume 113 / Number 12