curative pca radiotherapy

8/8/2019 Curative PCa Radiotherapy

1/5

Doctors and Patients Preferences for Participation andTreatment in Curative Prostate Cancer RadiotherapyPeep F.M. Stalmeier, Julia J. van Tol-Geerdink, Emile N.J.Th. van Lin, Erik Schimmel, Henk Huizenga,Willem A.J. van Daal, and Jan-Willem Leer

A B S T R A C T

PurposePhysicians hold opinions about unvoiced patient preferences, so-called substitute preferences.We studied whether doctors can predict preferences of patients supported with a decision aid.

MethodsA total of 150 patients with prostate cancer facing radiotherapy were included. After the initialconsultation, without discussing any treatment choice, physicians gave substitute judgments forpatients decision-making and radiation dose preferences. Physicians knew that several weeks later,

patients would be empowered by a decision aid supporting a choice between two radiation dosesinvolving a trade-off between disease-free survival and adverse effects. Subsequently, patientpreferences for decision making (whether or not they wanted to choose a radiation dose) and fortreatment (low or high dose) were obtained. The chosen radiation dose actually was administered.

ResultsOf the patients studied, 79% chose a treatment; physicians believed that 66% of the patientswanted to choose. Agreement was poor (64%; 0.13; P .11), and was better as patientsbecame more hopeful (odds ratio [OR] 4.4 per unit; P .001) and as physicians experienceincreased (OR 1.09 per year; P .02). Twenty percent of physicians preferences, 51% ofphysicians substitute preferences, and 71% of patients preferences favored the lower dose;agreement was again poor (70%; 0.2; P .03).

ConclusionPhysicians had problems predicting the preferences of patients empowered with a decision aid.

They slightly underestimated patients decision-making preferences, and underestimated patientspreferences for the less toxic treatment. Counseling might be improved by first informingpatientspossibly using a decision aidbefore discussing patient preferences.

J Clin Oncol 25:3096-3100. 2007 by American Society of Clinical Oncology

INTRODUCTION

When patients have unvoiced preferences, the phy-

sician decides in the patients best interest. To make

thisdecision, physicians maytryto imagineor judge

theunvoiced preferencesof patients. These so-called

substitute judgments have been studied in end-of-life decision making. A general finding is that agree-

ment between substitute and patient preferences is

poor,1-5 suggesting that patients needs and stan-

dards for informed consent may not be met.

Other preferences such as preferences for deci-

sion making and treatment selection have been stud-

ied. Decision-making preferences reflect whether or

not the patient wants to choose himself, or leave

the decision to the physician. Treatment prefer-

ences reflect which medical treatment the patient

wants. During the consultation, physicians hold

opinions or substitute judgmentsabout particular

patient preferences.How are these opinions or substitute judg-

ments formed? It is known that physicians may usetheir own preferences to form their substitute judg-

ments.1,4 It is also likely that substitute judgments

are formed by readily available patient characteris-

tics such as age, education, and disease severity. Forexample, patients decision-making preferences are

positively associated with female sex and higher ed-ucation,anddeclinewithageand disease severity.6-9

It is plausible thatthese factorsaffect substitute pref-

erences held by physicians.Substitute preferences are relevant for medical

decision making. They are likely to steer the infor-mationdetailandlevel of involvementthat isoffered

to patients. In clinical practice, physicians may as-

sume that they are a good judge of patients prefer-ences and preferred roles for decision making are

From the Departments of Radiation

Oncology and Medical Technology

Assessment, Radboud University

Nijmegen Medical Center, Nijmegen;

and Arnhems Radiotherapeutic Insti-

tute, Arnhem, the Netherlands.

Submitted May 24, 2006; accepted

April 20, 2007.

Supported in part by a grant from the

Dutch Cancer Society, Amsterdam, the

Netherlands (Project No. KUN 2001-2379

and KUN 2005-3457). The funding agree-

ment ensured the authors independence

in designing the study, interpreting the

data, and writing and publishing the re-

port. P.F.M.S. and J.J.V.T.-G. are sup-

ported by the sponsor.

Presented at the 10th Biennial Euro-

pean Meeting of the Society for Medi-

cal Decision Making, June 11-13, 2006,

Birmingham, United Kingdom.

Authors disclosures of potential con-

flicts of interest and author contribu-

tions are found at the end of this

article.

Clinical Trials Registry ISRCTN97145188.

Address reprint requests to Peep F.M.

Stalmeier, PhD, Radboud University

Nijmegen Medical Centre, MTA 138, PO

Box 9101, 6500 HB Nijmegen, the Neth-

erlands; e-mail: [email protected].

2007 by American Society of Clinical

Oncology

0732-183X/07/2521-3096/$20.00

DOI: 10.1200/JCO.2006.07.4955

JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T

V OL UM E 2 5 N UM BE R 2 1 J UL Y 2 0 2 00 7

3096

Copyright 2007 by the American Society of Clinical Oncology. All rights reserved.Information downloaded from jco.ascopubs.org and provided by ROMANELLO FOUNDATION on July 20, 2008 from .


2/5

notusuallyasked.10,11Patients,inturn,askfewquestions.12Hence,the

deliveryof informationis largelyphysician driven.13,14 Theamount ofinformation providedin practice is variable and often too incomplete

to facilitate patient decision making.14,15

The variation in information suggests that physicians know inadvance which patients want and which patients reject additional

involvement after being informed further. Our study question is

whether physicians indeed know these preferences. Given that thephysician judgment about patients involvement wishes precedes the

provision of additional information, physicians opinions were ob-tained before additional information was provided to patients. The

preferencesof patientswere obtainedafteradditional information was

providedbecausethe physician judgeswhatthe patient wants after thepatient is informed further. In addition, factors affecting agreement

between substitute and patient preferences were sought. The context

was realistic; that is, the patients choice was carried out.

METHODS

PatientsBetween June 2003 andFebruary 2005, all patients with a primary pros-

tate tumor (T1-3,N0, M0) scheduled to undergo radiotherapy were includedin this study.16,17 Exclusion criteria were mental disorders and insufficientknowledgeof theDutch language.Patients were enrolledin two locations: theRadboud University Nijmegen Medical Centre (Nijmegen, the Netherlands)and the Arnhems Radiotherapeutic Institute (Arnhem, the Netherlands).

ProcedureThe study was approved by the research ethics committees of both

hospitals. In thefirst visit to theclinic, which lastedabout 30to 45minutes,theradiationoncologist informedeligiblepatientsthat thisstudyfocusedon howto involve the opinion of patients in the treatment. The radiation oncologistthen asked patients if the researcher could contact them by phone about thisstudy. Physicians did not discuss the choice between two radiation doses with

the patient. Physicianscompletedinformation abouttheir (substitute) prefer-ences after the patient left the clinic. In the phone call, the researcher toldpatients that data were collected by means of an interview and several ques-tionnaires. Patients who agreed to participate were sent a consent form and abaseline questionnaire to be completed within a few days.

In the second visit, on average 19 days after the first visit, the researcherinterviewed the patients. In this interview, additional information was pro-vided with a decision aid. After 2 days, decision-making and treatment pref-erences were confirmed by telephone and noted in the patients medicalrecord. These latter responses were used for the analyses described in thisarticle. Subsequently, the preferred treatment (high- or low-dose radia-tion) was delivered. Physicians knew that patients preferences were to beobtained after receiving the decision aid; physicians were familiar with thecontent of the decision aid.

Interview With Decision AidIn the second visit, patients were told about the possibility of two treat-

ments in a semistructured interview. A decision aid explained the trade-offbetween the risks and benefits of higher or lower radiation dose.16,17 A higherdose leads to better (disease free) survival, but also to more adverse effects.Patients received outcome and risk information on the two alternative treat-ment options of 70 and 74 Gy effective radiation dose. A literature study

yielded data on the following outcomes18: 5-year overall survival, 5-yeardisease-free survival, severe erectile dysfunction, severe late GI adverse effects,and severe genitourinary adverse effects. Severe adverse effects weredefined asadverse effects that have an impact on daily activities. The probability that theseoutcomesoccurred (riskinformation)was presented subsequentlyin frequencies(xof100patients)andbymeansofpiecharts. 16Theoutcomeandriskinformationwasalso givento thepatients in writing and could be takenhome.

The information was tailored to the patient characteristics in terms ofprognostic risk and age category. Four separate information groups weredistinguished. Thefirst group consists of low-risk patients, whoare character-ized by a prostate-specific antigen (PSA) value less than 10 ng/mL, a Gleasonscore lessthan 7,and a T statusof T1 or T2. The other patients weredividedintoage categories of younger than 57.5 years, 57.5 to 72.5 years, and olderthan 72.5 years. After receiving the decision aid, 10% of the patientsrequested and received an additional consultation with the physician todiscuss the choice.

MeasuresData were collected on several variables that were expected to affect

patients preferences. Most patient data were collected after the first visit.Patient decision-making and treatment preferences were collected in the sec-ondvisit,during which thedecisionaid was presented. Foreach physician, werecorded sex, years of training as a radiation oncologist, and the number ofpatients included in the study.

Preferences of CliniciansSubstitute preferences were obtained at the end of the first visit; that is,

before patients were informed with the decision aid. Physicians were familiarwith the decision aid. With this background knowledge, a general instructionasked physicians to take the patient perspective. Three measures were ob-tained. The substitute decision-making preference was asked as follows: Do

you believe that this patient wants to decide for himself between the low andhigh-dose(yes/no)? Thesubstitutetreatment preference wasasked as follows:Assuming that the patient wants to choose, which dose do you believe that thepatientwillchoose (thelow dose/thehigh dose)? Thephysiciansownpreferencefortreatmentwasaskedasfollows:Accordingtoyourown opinion,whichdose ismostsuited for thispatient(thelow dose/thehigh dose/nopreference)?

Patient VariablesPreferences in patients. After receiving the decision aid, the patient

answered the decision-making preference question: Do you want to chooseone of the two treatment options, or do you want to leave the decision to thephysician (choose/leave/dont know)? The patient treatment preference wasasked as follows: Which dose do you prefer (the low dose/the high dose)?Responses were confirmed 3 days later by telephone; for the decision-makingquestion, the final response options were choose or leave.

Demographic variables and tumor characteristics. We collected self-report dataon demographic variables (age, marital status, having [grand]chil-dren, working status, education, and religion). Tumor characteristics (Tstatus, pretreatment PSA value, and Gleason score) were extracted from themedical records.

Psychological and well-being variables. The Problem-Solving Decision-Making Scale19 assessed general decision-making preferences using two ques-tions:Whenthe risks andbenefitsof radiotherapy areknownto you, (1)whodecides howacceptablethose risks andbenefits areto you, and(2) whodecideson the choice? The response scale ranged from the doctor alone (1), to Ialone (5). The second question was also analyzed separately and used as abaseline decision-making preference. The personality traits autonomy andconscientiousness were assessed with five itemseachfroma shortenedversionof a personality assessment instrument.20

General health in the previous week was assessed with an 11-point

horizontal rating scale ranging from worst imaginable (0) to best imaginablehealth state (10). Hopelessness, avoidance, and fighting spirit were assessedwith the Mental Adjustment to Cancer scale.21 Cancer worries were assessedwith three questions: Did you think of prostate cancer last week, did thesethoughts affect your mood, and did these thoughts affect your daily activi-ties?22,23 Data were obtained on anxiety and depression by means of theHospital Anxiety and Depression Scale.24 Prostate-specific quality of life wasassessed by means of the European Organisation for Research and Treatmentof Cancer Quality of Life Questionnaire PR25 prostate cancer module.25 Itcontains the following scales: urinary symptoms, bowel symptoms, treatmentrelated adverse effects (bloatedness, hot flashes, edema, weight gain/loss), andsexual functioning.

Knowledge. We asked patients to rate their knowledge (subjectiveknowledge) on prostate cancer and on the advantages and disadvantages of

Preference Prediction in Prostate Cancer RT

www.jco.org 3097



3/5

radiotherapy on a 10-point scale (ranging from very poor to excellent).Numeracy (ie, the ability to handle basic probability concepts) was assessedwith three questions (eg, convert 1 in 1,000 to a percentage).26

Information. Patients were asked to rate their preference for informa-tion on a scale ranging from 0 (I want to know nothing about my illness andtreatment) to 10 (I want to know as much as possible concerning my illnessand treatment). The patients perception of the amount of information pro-videdon prostatecancer andradiotherapywas ratedon a 7-pointscalerangingfrom far too little information to far too much information.

AnalysisAgreement between patients preferences and physicians substitute

preferences is examined for decision making and treatment preferences sepa-rately. Agreementcanarisefromchance;if onepredictsheadsallthe time,50%of coin tosses are predicted correctly. A measure correcting for chance agree-mentis the statistic, ranging from 0 to 1.27,28 Aof 0.2, 0.5, and0.8 indicatespoor,moderate, and good agreement, respectively. In bivariate and multivar-iate analyses, associations were sought between agreement and the physicianvariables (eg, years of training and number of patients contributed) and pa-tient variables (eg, demographic, medical, psychological, knowledge and in-formation) described above. In case of missing data, scale values werecalculated, if at least half of the items were filled out, by imputing the mean ofthe remaining items. For bivariate analyses using 2 tests, continuous datawere dichotomized using the median split. For PSA values, a cutoff of 10

ng/mL wasused; forGleasonscores, a cutoffof 7 wasused. In addition, anxietyand depression were dichotomized by use of a clinical cutoff point of 8.

Dichotimized variables associated with agreement at a level of P .20were included whenever possible as continuous variables in a hierarchicalmodel for binary outcomes (procedure GENMOD; SAS software,version 8.2;SAS Institute, Cary, NC) with agreement as the dependent variable. Thismodel accounted for patients clustering under physicians. Generalized esti-mating equations estimates were used to test for significance.

RESULTS

During the inclusion period, a total of 544 patients came for a

consultation. More than 60% of these patients were not eligiblebecause of local or distant metastases (n 121), previous radical

prostatectomy (n 94), avoidance of external-beam radiotherapy

(n 36), cognitive/mental problems (n 17), insufficient knowl-edge of the Dutch language (n 10), and other reasons (n 31).

The remaining 200 patients were asked to participate in the study.

One hundred fifty (75%) of these patients provided informedconsent and were included in the study. Patient characteristics are

categorized by informed consent in Table 1; the two groups did notdiffer. Physician characteristics are listed in Table 2. Because both

centers were training hospitals, physicians differed in experience and

number of patients seen for this study.

Decision-Making Preferences

The results fordecision-making preferences arelisted in Table 3.Substitutepreferencesfor decision making wereavailablein 142of 150physician-patient pairs. The overall agreement in Table 3 is 64%. The

agreement corrected for chance agreement, as expressed by, is 0.13

(P .11). This indicates a poor agreement. Hierarchical analyses

indicated that agreement for decision-making preferences improvedwith years of training of the radiation oncologist (odds ratio

[OR] 1.09 for each year; P .02); agreement improved as patientsbecame more hopeful (OR 4.4 for each unit change; P .001),

hopelessnessvaluesvaried between1 and3, andthepossible range was

0 to 3. In physicians with more than 6 years of experience, overallagreement improved but chance corrected agreement remained poor

( 0.26; P .02). Similarly, in more hopeful patients, overallagreement improved to 75% but chance corrected agreement re-mained poor ( 0.23; P .02). Agreement was also better inpatientswith a PSAvalue greater than 10 ng/mL (OR2.1; P .003)andin patientswithout bladder or bowel surgery (OR2.1; P .03).

Surprisingly, not even patients own baseline decision-makingpreference agreed with their finalchoice behavior (agreement64%; 0.13; P .1).

Treatment Preferences

Thirty-one patientsdidnot want to choose, thus treatment pref-erences were not recorded; hence, agreement with physicians substi-

tute preferences was undefined. In addition, physicians did not

Table 1. Patient Characteristics

Characteristic

Informed Consent

Yes (n 150) No (n 50)

Demographic items

Age, years

Mean 70 71

Standard deviation 6 5

Living with partner 88 NAChildren 94 NA

College education or more 36 NA

Religion or philosophy of life 78 NA

Medical variables

T status

T1 16 19

T2 36 38

T3 48 44

PSA, ng/mL

Mean 24 22

Standard deviation 31 23

Gleason score, mean 6.5 6.6

Low-risk status (ie, T1-2, PSA 10g/L, and Gleason score 7)

19 13

Hormone deprivation therapy 74 83

Location

Arnhem 49 40

Nijmegen 51 60

Abbreviations: NA, information not available; PSA, prostate-specific antigen.

Table 2. Physician Characteristics

Characteristic No.

Sex

Female 6Male 9

Years of experience as a physician

Mean 9

Standard deviation 7

Median 6

Range 1-21

Patients per physician

Mean 10

Standard deviation 11

Median 7

Range 1-42

Number of patients included in this study by a physician.

Stalmeier et al

3098 JOURNAL OF CLINICAL ONCOLOGY



4/5

provide substitute treatment preferences for 25 patients. As a result,

agreement was undefinedin 49 of 150patients, leaving 101physician-

patient pairs (Table 4). The overall agreement was 60% ( 0.20;P .03), indicating that physicians are able to predict beyond

chance which treatment the patient desires; however, the agreement

was poor. Agreement was equally poor for male and female physi-cians. Hierarchical models for treatment preference agreement were

not interpretable because the significance of determinants varied

strongly depending on the variables included in the model.

Physicians substitute treatment preferences were associatedstrongly with their own preferred treatment plan (2 10; P .001;

Table 5). Additional analyses (data not shown) showed that physi-cians own preferences are driven by the medical status of the patient,

such as higherT status. Male physicianswerefour times more likely to

prefer the higher dose (2 15; P .001).

DISCUSSION

Do physicians know the preferences of empowered patients?

Decision-making and treatment preferences were assessed in phy-sicians, and afterward in patients empowered by a decision aid.

Slightly more patients (79%) wanted to decide themselves than thenumber expected by physicians (66%). Physicians could not indi-cate reliably which patients wanted to choose. Patients could not

foresee their final preferences either. Regarding treatment prefer-

ences, patients favored the less toxic treatment (lower dose) morefrequently (71%) than expected by physicians (51%). Physicians

foresaw patients treatment preferences to some extent, but agree-ment was poor. Physicians own preference favored the lower dose

in only 20% of the patients. Strong points of this study are that a

concrete therapeutic choice was made by patients facing treatment,and that the treatment chosen was actually delivered.

The agreement between physician and patient preferences was

poor. This has been observed before.1-4,29-31 One could argue that

the low agreement arises because the clinicians did not discuss the

treatment choice with patients. This procedure, however, followedfrom our research rationale. In practice, physicians may vary infor-

mationor involvementusingtheir judgmentof patientpreferencesfor

decision making. The judgment about patient preferences is madebefore extra information is provided to patients. This study intended

to assess the accuracy of such prior judgments. Accordingly, patients

preferences were obtained after decision making was facilitated, and

physicians substitute preferences were measured before decisionmaking was facilitated.Agreement was not uniformly poor. In more experienced

physicians, agreement regarding decision making improved: this is

a new finding, and it provides evidence for the claim of olderphysiciansthat experience improves theirjudgments of patients. In

more hopeful patients, agreement also improved: this finding is

also new and deserves additional interpretation. The rates of pa-tient and substitute preferences in more or less hopeful patients

were considered, butcouldnot explain that agreementimproves inmore hopeful patients. An alternative explanation takes into ac-

count communication patterns in less hopeful patients. Less hope-

ful patients are more anxious and depressed. Such patients utter

more concerns, requests, and receive more information, directives,empathy, and longer visits.32-35 Hence, such patients are more

demanding,34 giving physicians less time to judge patient prefer-

ences, which in turn leads to a lower agreement.Other comments can be made. Physicians often believed they

were guessing about patients preferences. They would have pre-

ferred to use a dont know category while judging patientsdecision-making and treatment preferences. Dont know re-

sponses, unfortunately, can not be analyzed, leading to a loss of

valuable data. More importantly, by being forced to guess, someresidual agreement might still be detected. In any case, varying the

amount of information is a daily routine in clinical practice, soforcing a judgment is not unreasonable. One may argue further

that the decision aid is new for physicians, thus confusing theirjudgment. One should consider, however, that several physicianswere involved in the design of the decision aid. All physicians knew

the content of the decision aid.

These findingsraise concernsabout theprovision of informationand the incorporation of patient preferences in decision making.

There aretwoimplications. First, thefindingsillustratethatphysiciansneedhelpto determine patient preferences.Physicians can discuss the

preferences of the patient, but this is not always common practice.

In such a discussion, decision aids may improve agreement36 be-

cause they present decision-making information in a formatthat patients understand, thus helping patients to develop and state

Table 3. Substitute and Patient Preferences for Participation

Substitute Physician Preferences

Patient Preferences

Active Passive Total %

Active 77 16 93 66

Passive 35 14 49 34

Total 112 30 142

% 79 21

Table 4. Substitute and Patient Treatment Preferences


Patient Preferences

Low High Total

Low 42 10 52

High 30 19 49

Total 72 29 101

NOTE. Low and high refer to radiation intensity.

Table 5. Substitute and Physician Treatment Preferences

Physician Preferences


Low High Total %

Low 20 5 25 20

High 36 46 82 66

No preference 8 9 17 14

Total 64 60 124

% 52 48

NOTE. Low and high refer to radiation intensity.

Preference Prediction in Prostate Cancer RT

www.jco.org 3099



5/5

curative pca radiotherapy

Documents