|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2001 American Society for Clinical Oncology Patient Preference for Cancer Therapy: An Overview of Measurement ApproachesFrom the Department of Medical Decision Making, Leiden University Medical Center, Leiden, and Department of Medical Psychology, Academic Medical Center, University of Amsterdam, Amsterdam, the Netherlands. Address reprint to A.M. Stiggelbout, PhD, Medical Decision Making Unit, Leiden University Medical Center, PO Box 9600, 2300 RC Leiden, the Netherlands; email Stiggelbout{at}rullf2.medfac.leidenuniv.nl
PURPOSE: In the era of evidence-based medicine and shared decision making, the formal assessment of patient preference for treatments or treatment outcomes has attracted much attention. In this article, the two most common approaches to the evaluation of preference, ie, utility assessment and probability trade-off assessment, are described. The purpose is to provide clinicians with the background knowledge needed to interpret preference studies published in the literature and to judge whether the reported findings are relevant to their own patients. METHODS: An overview is given of the methods used to assess utilities and probability trade-off scores. Evidence on determinants of such scores is presented. Examples from oncology are provided. Because experience with the treatment plays an important role as a determinant of preferences for both treatments and treatment outcomes, special attention is paid to the interpretation of studies in the light of subject selection. Directions for future research are suggested. CONCLUSION: The choice of approach and the measuring instrument depend on the goal of the preference assessment. Normal psychologic processes, such as coping, adaptation, and cognitive dissonance reduction, cause patients who are about to undergo a therapy or have experienced a therapy to rate it more favorably than other patients do. This should be remembered when using evidence from the literature to inform patients or for patient decision making.
MANY THERAPEUTIC modalities for cancer not only have serious side effects but also have uncertain or limited benefits. In oncology, the benefit and toxicity of various approaches have to be weighed to determine the value of the therapy. Whether the improved survival due to a new treatment outweighs the side effects of that treatment, or whether a higher quality of life is to be preferred over a slightly improved survival, is a matter of value judgment. In other cases, in which the outcomes in terms of survival are equal for the different options involved, treatments may vary with respect to quality-of-life outcomes, and a weighing of these outcomes is required. When no treatment is superior with respect to both quality and length of life, or with respect to all dimensions of quality of life, descriptive information on the quality of life with and without treatment does not suffice. Descriptive quality-of-life instruments have been developed to assess the impact of disease and treatment on the various dimensions of functioning and well-being, and the resulting profile scores are based on the assumption that all symptoms of a given severity are equally significant to the patient. They have not been designed to incorporate the trade-offs between various dimensions of quality of life that are involved in decision making. When quality and length of life, or different dimensions of quality of life, have to be weighed, an evaluation of the health outcomes is required. In the clinical setting, such an evaluation process is usually performed in an informal and implicit manner. However, recent developments in shared decision making1 have created a need for a more explicit assessment of patient evaluations of treatment outcome. Furthermore, for the development of guidelines and the formulation of health care policy, formal evaluation of outcomes is part of what has become known as "evidence-based medicine." A steady increase has been seen in the number of studies on preference assessment, ie, the formal measurement of the strength of the preference of patients for a specific treatment2-4 or the outcome of such treatment.5,6 Readers without a background in preference research may find it difficult to interpret the results of such studies. The impact of subject selection on the findings presented may not be clear; the same applies for the moment of assessment (phase of the treatment) and the methods used to assess values. The primary purpose of this article is to familiarize a clinical audience with the methods used in preference assessment. This knowledge will help readers to evaluate the results obtained by different methods and to interpret (apparently) conflicting results. The second purpose of this article is to describe the demographic or clinical determinants of preference. Information on factors associated with preference may help clinicians to judge the applicability of study findings to their own patients. The first section of this article deals with the different levels of decision making in health care, as different approaches to preference assessment are used for these different levels. The most common assessment is the utility approach. This approach, the methods used, and the criteria for the selection of the appropriate instruments are described in the second section. The probability trade-off methods are discussed in the third section. The data on determinants of utilities and probability trade-off scores are evaluated in the fourth section. Whether the patient is about to start chemotherapy, is in the middle of, eg, the third course, or has successfully completed all courses may play a prominent role in his or her viewpoint on chemotherapy. Therefore, in the fifth section, the stability of patient preference is discussed.
Three levels of decision making are generally distinguished in health care.7,8 The different levels call for different sources of the preferences and for different measurement approaches. The first, the macro level, concerns decisions on allocation, or resource utilization. The perspective is that of society as a whole. Decisions about health care resources based on public interest call for the incorporation of society’s preferences in cost-effectiveness analyses. The concerns of an individual patient or for the health of that patient are generally of little interest when such decisions are required. This would suggest that these preferences should be obtained from a representative sample of members of the community.9-11 However, controversy exists on this matter, as discussed in depth in the report of the Panel on Cost-Effectiveness in Health and Medicine.9 It can be argued, eg, that persons experiencing a particular health state are better suited to provide an assessment of the value of that state. The public may harbor stereotypes and biases and thereby underestimate the value of a health state, not realizing that they would accommodate to suboptimal health. Indeed, people who have a disease or condition will generally value an associated health state higher than those who have not experienced it.9 However, the public may well reflect by their lower valuations the fact that people in suboptimal health are limited in their functioning and that it would still be worth striving for a minimization of disability and maximization of optimal health, even if empirical evidence shows that patients adapt to reduced health. Another argument against preferences from the general public relates to the fact that the judgments should be informed and competent. In some instances, eg, when complex outcomes are involved, there may therefore be practical reasons for obtaining preferences of individuals in a particular state of health rather than from members of the public whose level of understanding of the nature of the health states is not always accurate. The Panel also recommends that when there are important differences in preferences among subgroups, analysts should conduct sensitivity analyses to show the impact of these differences on the cost-effectiveness ratio.9 At the macro level, choices have to be made among programs in the case of limited resources. This implies that a generic approach to preference measurement is needed in which comparisons can be made across disease categories.10 Some methods for collecting preferences involve asking subjects directly for their preferences for health states, using techniques such as the standard gamble (SG) and the time trade-off (TTO) (both are explained in the next section). Another approach measures preferences indirectly, relying on health state classification systems, such as the Quality of Well-Being Scale,12 the Health Utility Index,13 and the EuroQoL.14 The latter approach is recommended by the Panel on Cost-Effectiveness in Health and Medicine9 because it obtains valuations from the general public for health outcomes in a standardized, generic way. Health state classification systems normally consist of two components: a descriptive system and a preference scoring formula. The descriptive systems cover a set of items pertaining to the domains of quality of life or functioning, such as mobility, cognitive functioning, and mood. A health state is described by indicating the patient’s level of functioning within each domain. (For instance, in the EuroQoL, each of five domains (mobility, self-care, usual activities, pain/discomfort, and anxiety/depression) is scored at three levels of severity, corresponding to "no problem," "some problem," and "extreme problem." Two additional states, "unconscious" and "dead," are also included. Thus, a total of 245 states (35 + 2) can be defined by this system.) Next, the scoring formula is used to calculate the preferences for the states thus defined by the descriptive system. The formula is based on preferences that have been obtained from the general public in earlier studies (partly by direct measurement, using methods such as SG and TTO, which are described in the next section, and partly by inference, either theoretical or statistical). Since premeasured preferences from samples of the general public are available for these systems, it suffices to have patients describe their health states in terms of the descriptive system and to use the scoring formula to obtain the preferences needed from the general public.10 A disadvantage of these generic systems is that they may lack the sensitivity to detect meaningful differences between relatively closely related outcomes. Increasingly, the descriptive systems of such indexes are used in clinical trials, to enable cost-effectiveness analyses from a societal perspective, using the preference weights from the general public. An in-depth discussion of preference assessment at the macro level is beyond the scope of this article, which pertains predominantly to clinical decision making. The interested reader is referred to the report of the Panel on Cost-Effectiveness in Health and Medicine.9 A more extensive discussion both of the choice of the respondents and of the classification systems can be found in this report. The second level, the meso level, pertains to clinical decision making for groups of patients with the same disease, with the development of guidelines as a specific case. The aim of decision making at the meso level is to decide on an optimal treatment policy for groups of patients with similar clinical characteristics. From this point of view, patient preference is relevant. The most common approach to preference measurement at this level is utility assessment. This approach assesses the value of treatment outcomes (ie, health states). The third, the micro level, applies to decision making for the individual patient. In this case, naturally, the patient’s own preference is assessed. The approach developed specifically for this level is the probability trade-off method, which evaluates treatment-outcome paths. This method often forms part of so-called shared decision-making programs. In the next two sections, these approaches are described.
PREFERENCES AT THE MACRO AND MESO LEVELS: UTILITY ASSESSMENT In a decision analysis, utilities are used as quality adjustment factors, to adjust a health state for its quality of life. Utilities for the outcomes are multiplied by the probabilities that these outcomes will occur to obtain the expected utility. The treatment with the highest expected utility is the preferred treatment. Decision analysis is thus based on expected utility theory, the normative theory for decision making under conditions of uncertainty (for more background, see, eg, Pliskin et al17 and Torrance18). In oncology, utilities and probabilities are commonly combined with length of survival in the respective states. The expected value of each treatment strategy is thus expressed in terms of quality-adjusted life years (QALYs). QALYs define health status in equivalents of "well-years" of life. They reflect the relative desirability of treatment outcomes with respect to quality of life and length of life. For example, 2 years in a state of health that is valued at only 50% of optimal health, ie, as having a utility of 0.5, are equivalent to 1 QALY (2 x 0.5 = 1). Figure 1 illustrates the decision tree for the decision (denoted by a square) between radiotherapy and surgery for 65-year-old patients with T3N0M0 laryngeal cancer.19 Radiotherapy has a 0.53 chance of locoregional cure, with "alive with natural speech" (utility = Un) as outcome. In the case of locoregional recurrence, there are two possible outcomes: a 0.57 chance of salvage after surgery, with "alive with artificial speech" (Ua) as an outcome, and a 0.43 chance of no salvage, which leads to early death (Ud; Ud = 0). Choosing surgery gives a 0.72 chance of "alive with artificial speech" and a 0.28 chance of locoregional recurrence, which is followed by a 0.38 chance of "alive with artificial speech" and a 0.62 chance of early death. In this example, the life expectancy after successful treatment is 14.1 years; after unsuccessful treatment 3 years, the utility of "alive with artificial speech" is 0.70 and the utility of "alive with natural speech" is 0.80.
The QALY calculation for surgery is as follows: (0.72 x 14.1 x 0.70) + (0.28 x 0.38 x 14.1 x 0.70) + (0.28 x 0.62 x 3 x 0.70) = 8.52 QALYs. In a similar way, the QALYs following radiotherapy can be obtained (9.13 QALYs).19 Since the number of QALYs for the radiotherapy strategy is higher than that for the surgery strategy, radiotherapy is preferred. A well-known application of QALYs in oncology is the quality-adjusted time without symptoms and toxicity (Q-TWiST) model.20,21 In this model, three health states are distinguished: (1) time spent with the subjective toxic effects of treatment (TOX); (2) time without symptoms of relapse or toxic effects of treatment (TWiST); and (3) time after disease relapse (REL). TOX and REL are weighted relative to TWiST by coefficients of utility. This model was devised particularly for the evaluation of quality-adjusted survival in chemotherapy trials.22,23 In studies based on the Q-TWiST method, values for the utilities are as a rule assumed rather than assessed directly, and subsequently varied in threshold analyses, also called sensitivity analyses. Such analyses result in combinations of utilities for TOX and REL (relative to TWiST) whereby one treatment strategy is superior to the other. The authors subsequently propose that clinicians assess these utilities with their patients to decide which treatment is superior.21 We do not know whether this model is applied in such a manner in daily clinical practice. Utility Assessment Methods: Description and Rationale for Selection The most commonly used methods to assess utilities are the SG, the TTO, and the visual analog scale (VAS). These methods are illustrated here by evaluating the state "alive with artificial speech" from Fig 1. In the SG method (Fig 2), a subject is offered the hypothetical choice between the sure outcome, A (living his remaining life expectancy in the state "alive, artificial speech"), and the gamble, B. The gamble has a probability p of the best possible outcome (optimal health, defined as 1) and a probability (1 - p) of the worst possible outcome (usually immediate death, defined as 0). By varying p, the value at which the subject is indifferent to the choice between the sure outcome and the gamble is obtained. The utility for the sure outcome, the state "alive, artificial speech," is equal to the value of p at the point of indifference (U = p x 1 + [1 - p] x 0 = p). Thus, if the subject is indifferent to the choice between his remaining life in "alive, artificial speech" and a gamble with a probability of 0.90 that his remaining life will be in optimal health and a probability of 0.10 of immediate death, the utility for "alive, artificial speech" is 0.90 (0.90 x 1 + 0.10 x 0).
In the TTO method (Fig 3), the subject is asked to choose between his remaining life expectancy in the state "alive, artificial speech" and a shorter life span in normal health. In other words, he is asked whether he would be willing to trade years of his remaining life expectancy to avoid artificial speech. As an example, a 65-year-old man is asked how many years x in a state of optimal health he considers equivalent to a period of 15 years (his remaining life expectancy) in the state "alive, artificial speech." By varying the duration of x, the point is found where he is indifferent to the choice between the two options. The simplest and most common way to transform this optimal-health equivalent x into a utility (ranging from 0 to 1) is to divide x by 15.
A VAS is a rating scale, a simple method that can be self-administered and therefore is often used to obtain evaluations of health states (Fig 4). The subject is asked to rate the state by placing a mark on a 100-mm horizontal or vertical line, anchored by optimal health and death (or sometimes by best possible health and worst possible health). The score is the number of millimeters from the "death" anchor to the mark, divided by 100. The VAS does not reflect any trade-off that a subject may be willing to make in order to obtain better health, neither in terms of risk nor in years of life.24
The SG, TTO, and VAS methods have all been recommended by some authors.25 Selection of an instrument should, however, include consideration of its measurement properties: feasibility, reliability, validity, and responsiveness (or sensitivity to change). All three methods have been reported to be feasible. For the SG and TTO methods, an interview setting is recommended, although under certain circumstances surveys may be possible (see Albertsen et al26 for an example involving patients with prostate cancer). With respect to reliability, no method seems to be clearly superior to the others. The reliability of all three methods has been judged acceptable for decision making at the group level.7 Very little research has been carried out with respect to responsiveness. Some studies have assessed the stability of utilities in patients whose quality of life or functional status changed over time, with conflicting results (see below, under Stability of Preferences). No studies, however, have formally compared the SG, TTO, and VAS with respect to responsiveness. Therefore, the choice depends predominantly on aspects of validity. Initially, the SG was recommended as the gold standard, or criterion, with respect to validity,7 because it is based on the axioms of expected utility theory. Later, several authors challenged this status of the method, given biases in the elicitation procedure.27,28 With respect to face validity, the TTO is considered by some to be the most valid method, because the question it poses is most closely associated with the sort of health care choices that need to be made.25,27,29,30 This holds for many situations in oncology, in which trade-offs between quality of life and length of life are involved and immediate risk is not a consideration. A study that has actually measured TTO utilities for use in a decision analysis in oncology involved women at high risk for hereditary breast cancer.31 In this study, incorporating utilities into the analysis sharply reduced the benefit of prophylactic surgery. The reduction of the quality of life due to cancer is much stronger than that due to prophylactic surgery, but the reduction due to surgery lasts for a patient’s lifetime, whereas the women may not incur cancer for decades. They may experience a nearly normal quality of life until they do so. In other situations, in which immediate risk is involved, the most appropriate method with respect to face validity would be the SG. Examples are evaluation of the value of bone marrow transplantation32 and decisions about more or less extensive surgery for cancer of the head of the pancreas. A disadvantage of the SG is that the use of probabilities is cognitively complex for the subject and may lead to biased utilities due to probability distortion.28 Generally utilities are biased upward, due to risk aversion: people accept the gamble only at very small probabilities (1 - p) of immediate death. This bias also results in ceiling effects, with all health states clustering at the top of the scale. The method may therefore lack the sensitivity to detect differences between closely related health states. For these reasons, the SG nowadays is used less often for medical decision making than the TTO. The VAS would only be appropriate if neither risk nor trade-offs between quality of life and length of life are involved and if weighing different dimensions of quality of life is the only aim. An example would be the choice between bisphosphonates in prostate cancer metastasized to the bone, whereby one treatment may give better pain relief but induce more nausea. If trade-offs of length of life and risk of death are involved, the use of the VAS is not justifiable, because it does not reflect these trade-offs. Nevertheless, in practice it has been used in such situations, for reasons of feasibility. A VAS score will underestimate the utility when minor impairments of health are at stake. Many subjects who indicate a score of less than 1.00 on a VAS are unwilling to trade life years (in a TTO) or risk of death (in an SG).24,33 An analysis based on VAS scores will subsequently overestimate the benefit of alleviating the health problem under study. In a situation in which a decision results in consequences with respect to life expectancy, one should therefore transform the inconsequential VAS score into a TTO or an SG utility. Power models have been proposed for this purpose.34,35 This transformation applies only to averaged scores, because at the individual level too much unexplained variation in SG or TTO scores remains for the VAS to be a reasonable substitute. Some, though not all, authors have been able to replicate these models in their data.33,36 In conclusion, the selection of the most valid assessment method depends on the decision under consideration, eg, is risk an important dimension or not.30 A point can be made for the use of more than one method, when feasible in terms of subject burden. By combining the information from a number of measures, one can increase the validity of the generalization of the results relative to that obtained by using only one measure.37
PREFERENCES AT THE MICRO OR INDIVIDUAL PATIENT LEVEL: PROBABILITY TRADE-OFF METHODS For these reasons, the patient’s relative preferences for treatment-outcome paths are nowadays assessed at the individual level. Probability trade-off methods, sometimes also called treatment trade-off or treatment preference methods, are used for this purpose. The process, or treatment, through which the outcomes are obtained forms part of the assessment procedure. In contrast, in utility assessment the preference for the health outcome after treatment is evaluated independently of the way in which this outcome is obtained. Thus, the approach at the individual patient level may seem to be less reductionist and may conform more to the broad view that patients have of their actual situation. The patient is presented with two clinical options (eg, cisplatin and cyclophosphamide [plan A] or cisplatin and paclitaxel [plan B] following surgery in advanced epithelial ovarian cancer, as described by Elit et al43; Fig 5). For each option, the probabilities of benefits and side effects are described. The patient is asked to state a preference for a treatment. If she prefers the less toxic treatment (plan A), the interviewer systematically either increases the probability of benefit from the more toxic treatment (plan B) or reduces the probability of benefit from the less toxic treatment (and vice versa if the more toxic treatment is preferred). The particular aspects of the treatments that are altered in this way and the direction in which they are changed are decided beforehand, according to the clinical characteristics of the problem and the nature of the research question.39 For example, they may include increasing the probability of side effects of treatment or decreasing the risk of recurrence or the chance of survival. The relative strength of preference for a treatment is assessed by determining the patient’s willingness to accept side effects of that treatment or forego benefits of the alternative treatment. This general approach has been adapted specifically to a variety of treatment decisions.
The resulting preference scores apply only to the original decision problem, and only the strength of preference for one treatment relative to another is obtained. The methods were not developed from the realm of expected utility theory and have not been evaluated in terms of the assumptions of the theory. For formal decision analysis, they are therefore not suitable. However, they can be used for decision support because they are tailored to the clinical problem at hand, and a judgment task will reflect the real situation more than utility assessment methods do.44 Reliability has been assessed for some of the specific applications of the trade-off technique and was found to vary widely from moderate44,45 to high.46,47 These methods seem to provide a promising way to help patients who wish to participate in the decision-making process clarify and communicate their values. They have indeed been used "at the bedside," with decision boards as visual aids, for adjuvant chemotherapy for breast cancer48 and ovarian cancer,43 radiotherapy for breast cancer,49 and chemotherapy for lung cancer.3 In a study in patients with ovarian cancer, Elit et al43 showed that survival information could be discussed with 12 out of 13 patients who took part in the testing of the instrument, whereas a pilot study had indicated that, in general, such data were not provided by oncologists. The decision board used in this study is shown in Fig 5. Whelan et al49 found that significantly more patients who used a decision board felt that they were actually offered a choice regarding breast irradiation compared with patients who did not use such an instrument. These results suggest that by presenting patients with information and by offering them a clear choice, a decision board may empower patients in a medical encounter49 and may help them to make decisions that are consistent with their personal values.50
DETERMINANTS OF PREFERENCES Clinical and Sociodemographic Factors A difference is seen between utilities and probability trade-off scores in the extent to which the variation among subjects is explained by clinical or sociodemographic factors. There is no compelling evidence that sociodemographic characteristics, such as age, sex, socioeconomic status, marital status, race, and religion, are associated with health state utilities.6,53-56 In general, the differences in evaluation that are attributable to personal characteristics are trivial compared with the differences that are due to different methods of assessment and framing of the questions.52 Probability trade-off scores, on the contrary, such as the minimally required benefit from treatment or the risk at which an adjuvant treatment is accepted, have been associated with age. Older patients generally prefer a less aggressive approach to treatment than younger patients do.57-60 Positive social well-being and children living at home ("someone to live for") have been found to predict patients’ willingness to accept aggressive treatment in hypothetical vignettes.57,61 Cullen et al2 found that patients with testicular cancer who did not have children chose surveillance over adjuvant chemotherapy up to higher levels of the recurrence risk compared with patients who already had a family. This may refer to the same phenomenon of "someone to live for," but in this particular study, fear of infertility may also have played a role in the preference for surveillance among men without children. Experience With the Health State or the Treatment Subjects who have experience with a health state are generally found to assign higher utilities to that state than subjects less familiar with the state, such as health care workers or members of the general public.9 In general, patients with experience seem to assign the highest values, members of the general public or patients without experience assign the lowest, and health care professionals assign values in between.6,9,56,62-65 Several explanations for this phenomenon have been offered. Health care professionals and patients are better informed about health outcomes than members of the general public and may therefore perceive these outcomes as less threatening. Moreover, someone who has been in a particular state of health for some time may perceive that state as being more desirable than someone who has not experienced it, given normal psychologic processes, such as coping and adaptation. Finally, a patient facing a life-threatening illness may be more concerned about the risk of immediate death because he or she perceives it as less hypothetical than someone who is not in immediate danger of dying. Thus, the patient will assign a higher utility on an SG than a healthy subject. There is some indication that subjects who judge their own health as poor place higher values on all adverse health states, a so-called valuation shift.66 This is in line with the finding that subjects in a certain health state generally rate that state higher than subjects who are not in that state. Because the study in which this valuation shift was observed used a VAS and not an SG, adaptation may be a more likely explanation than the described reluctance to take the risk of immediate death. Patients who are about to undergo treatment or have experienced a treatment generally have also been found to have a stronger preference for that treatment in terms of probability trade-off scores than both patients without experience and health care professionals.2,3,5,59,67 The explanation generally given for this phenomenon is the psychologic process called cognitive dissonance reduction. Subjects make their preferences agree with the decision that was made and they remain consonant even after experiencing toxicity.45 This finding is in line with evidence that subjects who have experience with a health state assign higher utilities to that state. In evaluating a study, it is therefore important to note whether the subjects in the study have already experienced the treatment or, if not, whether the decision to undergo treatment has already been made. In these cases, the preferences may differ from those that would have been obtained before the decision-making process and may be the result of psychologic adaptation mechanisms. Moreover, studies to assess preferences among patients who have undergone treatment generally do not include patients with a poor outcome despite the treatment. One can expect the attitude of patients toward adjuvant therapy to be more negative once the tumor has metastasized than when the patient is still disease-free. This has to be established empirically, though.
STABILITY OF PREFERENCES: IMPLICATIONS FOR THE DESIGN AND INTERPRETATION OF STUDIES Although results for less severe states have been found to be stable at the group level (ie, stable means), large intraindividual variation may still be seen (low correlation coefficients). Such figures are seldom presented, however. Should preferences be assessed from patients who have experienced the treatment or from those who are about to undergo treatment? For QALY calculations for decision making at the meso level, the preferences or utilities of patients in the actual health states involved are needed. For decision making for the individual patient, the preference beforehand seems to be relevant. However, it may still be useful for clinicians to know the preferences of patients who have experienced the treatment, so that the physician can explain to the patient that preferences may change during the course of treatment. The patient can then take into account the fact that his or her valuation may change over time, generally in a direction more favorably inclined toward therapy.
Utility assessment is commonly used to assess health state preferences for medical decision making. It is applied predominantly for guideline development and policy making, that is, for decision making for patient groups. For decisions for the individual patient, the methods are not sufficiently reliable, but they may help a patient clarify his or her own values. An alternative to utility assessment is the probability trade-off technique, which results in a preference score for one treatment relative to another. It is not utility-based and is therefore not suitable for formal decision analysis. Since the task resembles the real-life decision problem at hand, it may be more relevant for the patient than the methods for utility assessment. Formal evaluation of probability trade-off methods, preferably in a controlled design, is needed to assess whether they have added value in the decision-making process. Large interindividual variation is generally seen in utilities, which is not easily explained by patient characteristics. More in-depth studies are needed to explore this variation and to obtain insight into the way in which preferences are constructed.71 Relatively little is known about the stability of preferences, especially for severe outcome states. Moreover, although results for less severe states have been found to be stable at the group level, large intraindividual variation may still exist. The finding that patients who have experienced a treatment value such treatment higher than patients who have not may indeed indicate that preference shifts occur after treatment, possibly motivated by psychologic adaptation processes. Insight into the psychologic factors that play a role in positive reappraisal of treatments once experienced may be of help to patients who have difficulty coping with the treatment.
The authors thank W.R. ten Hove for his critical reading of an earlier version of the manuscript, P. Krabbe for the illustration used in Fig 1, and two anonymous reviewers for their very constructive comments.
1. Charles C, Gafni A, Whelan T: Decision making in the physician-patient encounter: Revisiting the shared treatment decision-making model. Soc Sci Med 49: 651-661, 1999 2. Cullen MH, Billingham LJ, Cook J, et al: Management preferences in stage I non-seminomatous germ cell tumours of the testis: An investigation among patients, controls and oncologists. Br J Cancer 74: 1487-1591, 1996[Medline]
3.
Brundage MD, Davidson JR, Mackillop WJ: Trading treatment toxicity for survival in locally advanced non-small cell lung cancer. J Clin Oncol 15: 330-340, 1997
4.
Lindley C, Vasa S, Sawyer WT, et al: Quality of life and preferences for treatment following systemic adjuvant therapy for early-stage breast cancer. J Clin Oncol 16: 1380-1387, 1998 5. Stiggelbout AM, Kiebert GM, De Haes JCJM, et al: Surveillance versus adjuvant chemotherapy in stage I non-seminomatous testicular cancer: A decision analysis. Eur J Cancer 32A: 2267-2274, 1996
6.
Hayman JA, Fairclough DL, Harris JR, et al: Patient preferences concerning the trade-off between the risks and benefits of routine radiation therapy after conservative surgery for early-stage breast cancer. J Clin Oncol 15: 1252-1260, 1997 7. Torrance GW: Measurement of health state utilities for economic appraisal: A review. J Health Econ 5: 1-31, 1986[Medline] 8. Sutherland HJ, Till JE: Quality of life assessments and levels of decision making: Differentiating objectives. Qual Life Res 2: 297-303, 1993[Medline] 9. Gold MR, Siegel JE, Russel LB, et al: Cost-Effectiveness in Health and Medicine. New York, NY, Oxford University Press, 1996
10.
Russel LB, Gold MR, Siegel JE, et al: The role of cost-effectiveness analysis in health and medicine: Panel on Cost-effectiveness in Health and Medicine. JAMA 276: 1172-1177, 1996
11.
Weinstein MC, Siegel JE, Gold MR, et al: Recommendations of the Panel on Cost-Effectiveness in Health and Medicine. JAMA 276: 1253-1258, 1996 12. Kaplan RM, Bush JW, Berry CC: Health status index: Category rating versus magnitude estimation for measuring levels of well-being. Med Care 17: 501-525, 1979[Medline] 13. Torrance GW, Furlong WJ, Feeny DH, et al: Multi-attribute preference functions: Health Utilities Index. Pharmacoeconomics 7: 503-520, 1995[Medline] 14. The EuroQoL Group: EuroQoL: A new facility for the measurement of health-related quality of life. Health Policy 16:199-208, 1990 15. Bush JW: Relative preference versus relative frequencies in health-related quality of life evaluations, in Wenger NK, Mattson ME, Furberg CD (eds): Assessment of Quality of Life in Clinical Trials of Cardiovascular Therapies. New York, NY, Le Jacq, 1984, pp 118-139 16. Torrance GW, Feeny D: Utilities and quality-adjusted life years. Int J Technol Assess Health Care 5: 559-575, 1989[Medline] 17. Pliskin JS, Shepard DS, Weinstein MC: Utility functions for life years and health status. Oper Res 28: 206-224, 1980 18. Torrance GW: Utility approach to measuring health-related quality of life. J Chron Dis 40: 593-600, 1987[Medline]
19.
Van der Donk J, Levendag PC, Kuijpers AJ, et al: Patient participation in clinical decision-making for treatment of T3 laryngeal cancer: A comparison of state and process utilities. J Clin Oncol 13: 2369-2378, 1995 20. Goldhirsch A, Gelber RD, Simes RJ, et al: Costs and benefits of adjuvant therapy in breast cancer: A quality-adjusted survival analysis. J Clin Oncol 7: 36-44, 1989[Abstract] 21. Gelber RD, Cole BF, Goldhirsch A, et al: Adjuvant chemotherapy plus tamoxifen compared with tamoxifen alone for postmenopausal breast cancer: Meta-analysis of quality-adjusted survival. Lancet 347: 1066-1071, 1996[Medline]
22.
Cole BF, Gelber RD, Kirkwood JM, et al: Quality-of-life-adjusted survival analysis of interferon alfa-2b adjuvant treatment of high-risk resected cutaneous melanoma: An Eastern Cooperative Oncology Group study. J Clin Oncol 14: 2666-2673, 1996
23.
Gelber RD, Goldhirsch A, Cole BF, et al: A quality-adjusted time without symptoms or toxicity (Q-TWiST) analysis of adjuvant radiation therapy and chemotherapy for resectable rectal cancer. J Natl Cancer Inst 88: 1039-1045, 1996 24. Robinson A, Dolan P, Williams A: Valuing health status using VAS and TTO: What lies behind the numbers? Soc Sci Med 45: 1289-1297, 1997 25. Nord E: Methods for quality adjustment of life years. Soc Sci Med 34: 559-569, 1992 26. Albertsen PC, Nease RF, Potosky AL: Assessment of patient preferences among men with prostate cancer. J Urol 159: 158-163, 1998[Medline] 27. Richardson J: Cost utility analysis: What should be measured—Utility, value or healthy year equivalents? Soc Sci Med 39: 7-21, 1995 28. Wakker P, Stiggelbout A: Explaining distortions in utility elicitation through the rank-dependent model for risky choices. Med Decis Making 15: 180-186, 1995 29. Gerard K, Dobson M, Hall J: Framing and labelling effects in health descriptions: Quality adjusted life years for treatment of breast cancer. J Clin Epidemiol 46: 77-84, 1993[Medline] 30. De Haes JCJM, Stiggelbout AM: Assessment of values, utilities and preferences in cancer patients. Cancer Treat Rev 22: 13-26, 1996 (suppl A) 31. Grann VR, Panageas KS, Whang W, et al: Decision analysis of prophylactic mastectomy and oophorectomy in BRCA1-positive or BRCA2-positive patients. J Clin Oncol 16: 979-985, 1998[Abstract]
32.
Hillner BE, Smith TJ, Desch CE: Efficacy and cost-effectiveness of autologous bone marrow transplantation in metastatic breast cancer: Estimates using decision analysis while awaiting clinical trial results. JAMA 267: 2055-2061, 1992 33. Stiggelbout AM, Eijkemans MJC, Kiebert GM, et al: The "utility" of the visual analog scale in medical decision making and technology assessment: Is it an alternative to the time trade-off? Int J Technol Assess Health Care 12: 291-298, 1996[Medline] 34. Torrance GW: Social preferences for health states: An empirical evaluation of three measurement techniques. Socio-Econ Plan Sci 10: 129-136, 1976 35. Torrance GW, Feeny DH, Furlong WJ, et al: Multiattribute utility function for a comprehensive health status classification system: Health Utilities Index Mark 2. Med Care 34: 702-722, 1996[Medline]
36.
Bosch JL, Hunink MGM: The relationship between descriptive and valuational quality-of-life measures in patients with intermittent claudication. Med Decis Making 16: 217-225, 1996 37. Nunnally JC, Bernstein IH: Psychometric Theory. New York, NY, McGraw-Hill Inc, 1994
38.
Llewellyn-Thomas HA, Williams JI, Levy L, et al: Using a trade-off technique to assess patients’ treatment preferences for benign prostatic hyperplasia. Med Decis Making 16: 262-272, 1996 39. Llewellyn-Thomas HA: Investigating patients’ preferences for different treatment options. Can J Nursing Res 29: 45-64, 1997
40.
Jansen SJT, Stiggelbout AM, Wakker PP, et al: Patient utilities for cancer treatments: A study on the chained procedure for the standard gamble and time trade-off. Med Decis Making 18: 391-399, 1998 41. Johnston K, Brown J, Gerard K, et al: Valuing temporary and chronic health states associated with breast screening. Soc Sci Med 47: 213-222, 1998 42. Unic I: Patient’s preferences in individual treatment selection for patients at high risk of breast cancer. Nijmegen, the Netherlands, University of Nijmegen, 1999 [thesis] 43. Elit LM, Levine MN, Gafni A, et al: Patients’ preferences for therapy in advanced epithelial ovarian cancer: Development, testing, and application of a bedside decision instrument. Gynecol Oncol 62: 329-335, 1996[Medline] 44. Kiebert GM, Stiggelbout AM, Leer JWH, et al: Test-retest reliabilities of two treatment preference instruments in measuring utilities. Med Decis Making 13: 133-140, 1993 45. O’Connor AMC, Boyd NF, Warde P, et al: Eliciting preferences for alternative drug therapy in oncology: Influence of treatment outcome description, elicitation technique and treatment experience on preferences. J Chron Dis 40: 811-818, 1987[Medline] 46. Sebban C, Browman G, Gafni A, et al: Design and validation of a bedside decision instrument to elicit a patient’s preference concerning allogenic bone marrow transplantation in chronic myeloid leukemia. Am J Hematol 48: 221-227, 1995[Medline]
47.
Brundage MD, Davidson JR, Mackillop WJ, et al: Using a treatment-tradeoff method to elicit preferences for the treatment of locally advanced non-small-cell lung cancer. Med Decis Making 18: 256-267, 1998 48. Levine MN, Gafni A, Markham B, et al: A bedside decision instrument to elicit a patient’s preference concerning adjuvant chemotherapy for breast cancer. Ann Int Med 117: 53-58, 1992 49. Whelan TJ, Levine MN, Gafni A, et al: Breast irradiation postlumpectomy: Development and evaluation of a decision instrument. J Clin Oncol 13: 847-853, 1995[Abstract]
50.
Molenaar S, Sprangers MA, Postma-Schuit FC, et al: Feasibility and effects of decision aids. Med Decis Making 20: 112-127, 2000 51. McNeil BJ, Pauker SG, Sox HC, et al: On the elicitation of preferences for alternative therapies. N Engl J Med 306: 1259-1262, 1982[Abstract] 52. Froberg DG, Kane RL: Methodology for measuring health state preferences: III. Population and context effects. J Clin Epidemiol 42: 585-592, 1989[Medline]
53.
Weeks JC: Preferences of older cancer patients: Can you judge a book by its cover? J Natl Cancer Inst 86: 1743-1744, 1994 54. Tsevat J, Cook EF, Green ML, et al: Health values of the seriously ill. Ann Intern Med 122: 514-520, 1994
55.
Chen AY, Daley J, Thibault GE: Angina patients’ ratings of current health and health without angina: Associations with severity of angina and comorbidity. Med Decis Making 16: 169-177, 1996 56. Tsevat J, Solzan JG, Kuntz KM, et al: Health values of patients infected with human immunodeficiency virus: Relationship to mental health and physical functioning. Med Care 34: 44-57, 1996[Medline] 57. Bremnes RM, Andersen K, Wist EA: Cancer patients, doctors and nurses vary in their willingness to undertake cancer chemotherapy. Eur J Cancer 31A: 1955-1959, 1995 58. McQuellon RP, Muss HB, Hoffman SL, et al: Patient preferences for treatment of metastatic breast cancer: A study of women with early-stage breast cancer. J Clin Oncol 13: 858-868, 1995[Abstract] 59. Ludwig H, Fritz E, Neuda J, et al: Patient preferences for interferon alfa in multiple myeloma. J Clin Oncol 15: 1672-1679, 1997[Abstract]
60.
Silvestri G, Pritchard R, Welch HG: Preferences for chemotherapy in patients with advanced non-small cell lung cancer: Descriptive study based on scripted interviews. BMJ 317: 771-775, 1998 61. Yellen SB, Cella DF: Someone to live for: Social well-being, parenthood status, and decision-making in oncology. J Clin Oncol 13: 1255-1264, 1995[Abstract] 62. Sackett DL, Torrance GW: The utility of different health states as perceived by the general public. J Chron Dis 31: 697-704, 1978[Medline] 63. O’Connor AM: Effects of framing and level of probability on patients’ preferences for cancer chemotherapy. J Clin Epidemiol 42: 119-126, 1989[Medline] 64. Boyd NF, Sutherland HJ, Heasman KZ, et al: Whose utilities for decision analysis? Med Decis Making 10: 58-67, 1990 65. Ashby J, O’Hanlon M, Buxton MJ: The time trade-off technique: How do the valuations of breast cancer patients compare to those of other groups? Qual Life Res 3: 257-265, 1997 66. Kind P: The effect of past and present illness experience on the valuations of health states. Med Care 33: AS255-AS263, 1995 (suppl A)[Medline] 67. Slevin ML, Stubbs L, Plant HJ, et al: Attitudes to chemotherapy: Comparing views of patients with cancer with those of doctors, nurses, and general public. BMJ 300: 1458-1460, 1990 68. Llewellyn-Thomas HA, Sutherland HJ, Ciampi A, et al: The assessment of values in laryngeal cancer: Reliability of measurement methods. J Chronic Dis 37: 283-291, 1984[Medline] 69. Llewellyn-Thomas HA, Sutherland HJ, Thiel EC: Do patients’ evaluations of a future health state change when they actually enter that state? Med Care 31: 1002-1012, 1993[Medline]
70.
Jansen SJT, Stiggelbout AM, Wakker PP, et al: Unstable preferences: A shift in valuation or an effect of the elicitation procedure? Med Decis Making 20: 62-71, 2000 71. Slovic P: The construction of preference. Am Psychol 50: 364-371, 1995 Submitted August 25, 1999; accepted July 27, 2000.
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2001 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
) between radiation therapy and surgery for 65-year-old men with T3N0M0 laryngeal cancer. Probabilities of outcomes are displayed after each chance node (
. Adapted.
