Patients' preferences for the management of non-metastatic prostate cancer: discrete choice experiment
BMJ 2004; 328 doi: https://doi.org/10.1136/bmj.37972.497234.44 (Published 12 February 2004) Cite this as: BMJ 2004;328:382All rapid responses
Rapid responses are electronic comments to the editor. They enable our users to debate issues raised in articles published on bmj.com. A rapid response is first posted online. If you need the URL (web address) of an individual response, simply click on the response headline and copy the URL from the browser window. A proportion of responses will, after editing, be published online and in the print journal as letters, which are indexed in PubMed. Rapid responses are not indexed in PubMed and they are not journal articles. The BMJ reserves the right to remove responses which are being wilfully misrepresented as published articles or when it is brought to our attention that a response spreads misinformation.
From March 2022, the word limit for rapid responses will be 600 words not including references and author details. We will no longer post responses that exceed this limit.
The word limit for letters selected from posted responses remains 300 words.
The boundary conditions of the present study are well defined:
Prostate cancer has a majority age of diagnosis of 65-plus years and
average age of survival of 5-plus years [1].
However, the diagnosis of prostate cancer for a younger man can spell
an early death[4]. In the UK nearly 1000 men of working age die every
year from prostate cancer[2]. In the USA [3] the number of pre-retirement
deaths due to prostate cancer is about 5000. Treatment needs of this
small but significant population requires attention because the chioces
made will impact so much more on quantity and quality of life. Putting the
same questions used in this study of 70 year olds to a 50 year old man, is
quite a different matter.
Following diagnosis at young age, decline traces a familiar pattern
for these men[5]: Failing prostatic surgery or radiotherapy (or both) for
the more aggressive type of prostate cancer, chemical castration or long
term hormone block (HB) is generally prescribed (e.g. gonadorelin
analogues goserelin, leuprorelin or triptorelin )[6]. I prefer the term
"chemical castration" to the more euphemistic "hormone block", since this
treatment effectively simulates physical castration with accompanying
iatrogenic symptoms, the so-called androgen deficiency syndrome [7,8].
The younger man diagnosed with prostate cancer needs an entirely new
approach to treatment that will challenge current medical thinking.
References
1. WHO World Health Statistics Annual 1997; Geneva; ISBN 92 4 06
7960 X. p. B-636
2. ibid WHO World Health Statistics Annual 1997; p. B-435
3. ibid WHO World Health Statistics Annual 1997.
4. Dijkman GA, Debruyne FM. Epidemiology of prostate cancer. Eur
Urol. 1996;30(3):281-95. Review. PMID: 8931959. Figure 3 p. 283
5. Cancer Causes Control 2002 Jun;13(5):435-43 ; Effect of young age
on prostate cancer survival: a population based assessment (United States)
; Merril RM, Bird JS. PMID 12146848
6. Section 8.3.4.2 Prostate Cancer and Gonadorelin Analogues;
British National Formulary, March 2003; ISBN 0 85369 555 5
7. Strum SB, McDermed JE, Scholz MC, Johnson H, Tisman G. Anaemia
associated with androgen deprivation in patients with prostate cancer
receiving combined hormone blockade. Br J Urol. 1997 Jun;79(6):933-41.
PMID: 9202563
8. Strum SB, Scholz MC & McDermed JE: The Androgen Deprivation
Syndrome: the incidence and severity in prostate cancer patients
receiving hormone blockade. Proc Amer Soc Clin Oncol. 17: 316A, 1998.
Competing interests:
None declared
Competing interests: No competing interests
Discrete choice experiments (DCE), long used in marketing research
(Louviere et al. 2003), are gaining popularity as a means of eliciting
preferences for health services. If properly designed, such experiments
allow efficient and independent estimation of the effects of interest. In
the paper by Sculpher et al (2004), these effects are factors that may be
important in a patient's decision about which treatment to choose for non-
metastatic prostate cancer. While this is a useful and interesting
application of stated preference methods, we have some concerns about the
particular DCE used.
The structure of the options presented in each choice and the set of
all choice sets used in a DCE determine which effects can be estimated.
Sculpher et al (2004) identify 8 attributes that may influence a
patient’s decision making; 6 treatment side-effects plus survival benefit
and cost. They then divide these into two “parts”, effectively creating
two separate choice experiments. Each part has 5 attributes; survival and
cost appear in both parts, while 3 side-effects appear in one part but not
the other. This was done to “avoid overburdening patients with too many
attributes”. While we agree that the effect of task complexity on
respondent consistency may be an issue in some cases, dividing attributes
into separate choice experiments creates four (related) problems: 1) the
hypothetical choice fails to mimic the real choice that men face in
trading off across the full set of treatment side-effects; 2) information
about the correlation between estimates of attribute effects in the two
parts of the DCE cannot be obtained; 3) the variance of the random
components in the two parts may differ, which would affect estimates of
the attribute effects and inferences; and 4) it may introduce bias due to
omitted variables and/or context effects. Our concerns are based on a
great deal of research summarized in Louviere, Hensher and Swait (2000,
Chapter 8), and we are unaware of any theoretical or empirical work that
would suggest that this is good practice.
While it is good that the authors have chosen an orthogonal main
effects design, several important details remain unclear. Which main
effects plan was used? Which attribute levels were used in each version
for each part? For example, although the “out-of-pocket expenses”
attribute is described as having 16 levels, the authors state, “Each
version of the questionnaire presented different levels of the cost
attribute...” so we do not know how many levels were used in any
particular design. Finally, how were the pairs chosen? It is not possible
to assess the statistical efficiency of the design (Street and Burgess
2004) unless all the pairs used are displayed.
We hope that in the future the journal will require that authors
provide sufficient information about the DCEs used to better inform
readers’ judgment and interpretation of results and conclusions.
Louviere J, Street D and Burgess L (2003). A 20+ years retrospective
on choice experiments. In Y Wind and PE Green, ed. Marketing Research and
Modeling: Progress and Prospects, Kluwer, New York.
Louviere JJ, Hensher DA and Swait JD (2000). Stated choice methods:
analysis and applications Cambridge University Press, Cambridge, U.K. ;
New York.
Sculpher M, Bryan S, Fry P, de Winter P, Payne H and Emberton M
(2004). Patients' preferences for the management of non-metastatic
prostate cancer: discrete choice experiment. BMJ, 0, 379724972-379724970.
Street D and Burgess L (2004). Optimal and Near-Optimal Pairs for the
Estimation of Effects in 2-level Choice Experiments. Journal of
Statistical Planning and Inference, 118, 185-199.
Competing interests:
None declared
Competing interests: No competing interests
Editor—
I read with great interest the article by Sculpher et al (1), it is
indeed commendable. However, instructing their patients to assume a five
year survival may lead to a feeling of ‘reconciliation and palliation’ on
the part of the patient and hence bias them to choose the most comfortable
option knowing that 'death is near'. Hence, the study perhaps should not
have included the five-year survival assumption by the patient.
Particularly so with the absence of concrete evidence in the literature
that their grade and stage may put them in this specific survival time
frame.
In this study, it is unclear as to whether the patients were made
fully aware of the implications of their choice on the survival benefit
literature wise. Reziciner et al (2) showed in his study involving
recurrence of cancer of the prostate after initial treatment with
diethylstilbestrol (DES) in a homogeneous series, that practically all
well differentiated prostatic cancers, regardless of their stage,
responded remarkably well to first-line treatment with sufficient doses of
DES. When correctly monitored, practically none of these cancers escaped
and early stages of escape can be salvaged. It was also shown that true
escape occurs all the earlier and evolves all the more rapidly for
advanced, poorly differentiated cancers, but this is not constant. I am
interested how the knowledge of the current literature may have influenced
the patients’ decision in the study. They should have had at least a
briefing of the current choices in more detail than the average counselled
patient.
The therapeutic dilemma that patients and clinicians face is the
timing of androgen suppression and this study illustrates the difficulty
of choice. Until we better understand and conquer the management of
hormone escaped prostate cancer patients, the dilemma will continue to
conquer us. The choices that patients make are most important in our
management at present. Their choice, however, would be irrational if they
were not presented with a reasonable cache of facts about their disease
process.
Respectfully,
Sashi Kommu.
1)Sculpher M,Bryan S,Fry P,de Winter P,Payne H, Emberton M. Patients'
preferences for the management of non-metastatic prostate cancer: discrete
choice experiment. BMJ 2004; 328: 382-0.
(2)Reziciner S.Recurrence of cancer of the prostate after initial
treatment with diethylstilbestrol (DES) in a homogeneous series of 175
cases. Ann Urol (Paris) 1997;31(4):213-24.
Competing interests:
None declared
Competing interests: No competing interests
I note that the CT image on the cover of the BMJ of 14th February has
been coloured in. I have seen this often before in general practice
magazines, but I am surprised that it is felt necessary in a scientific
journal. Are CT scans not felt visually stimulating enough? To arbitrarily
add colour to a black and white dataset is essentially manipulating data
to fit the result one wishes to show. If the colours are intended to
introduce clarity then use of green and brown together does not aid
interpretation for the 10% of the male population who are colour blind. I
note that the image has also been incorrectly labeled as an MRI scan.
Future editions of the BMJ may benefit from the editiorial input of a
radiologist.
Competing interests:
None declared
Competing interests: No competing interests
Interesting, but hypothetical
Sir,
I found the paper by Dr Sculpher and collagues highly interesting and
relevant, but what strikes me is the hypothetical nature of the decision
making. I am afraid people tend to value things differently when they
actually experience the situation - which is the shortcoming of utility
assessment. A more relevant approach would be conducting a trial of actual
patients choosing the treatment for themselves. We have tried to adopt
such a strategy in a trial where the intervention was different level of
guidance for choosing the treatment (Auvinen et al. BJU Int 2004;93:52-
56). Patients who received a treatment recommendation were more frequently
treated surgically than those for whom no preference between treatment
options was given by the physician. This result is in accordance with the
findings by Sculpher and others. However, I feel more knowledge can be
gained by applying a rigoristic study design in research into patient
participation in decision-making ("empirical ethics").
Competing interests:
None declared
Competing interests: No competing interests