Medical Decision Making

The following questions describe the scope of this paper. When decision trees are used to analyze optimal decisions, should end nodes be evaluated on the basis of QALYs or on the basis of healthy-years equivalents? Which measures should be used in communications with others, e.g., patients? Which of these measures incorporate nsk attitudes, and which do not? It is demonstrated that the healthy-years equivalent measure does not stand scrutiny. Key words: utility; QALY; quality of life; healthy-years equivalent; standard gamble. (Med Decis Making 1996;16:207-214)

The study objective was to assess the relationship between descriptive and valuational quality-of-life measures in patients with intermittent claudication. In telephone inter views, 68 patients completed a questionnaire consisting of a descriptive health status measure (RAND 36-Item Health Survey 1.0), and several valuational measures (stan dard gamble, time tradeoff, rating scale, and McMaster health utility index). All mea sures demonstrated reduced quality of life in the patients. Scores on the RAND-36 dimensions correlated moderately well with the rating scale and McMaster health utility index (R = 0.37-0.67) but less well with the standard gamble and the time tradeoff (R = 0.10-0.46). Multiple regression analysis demonstrated that 28% of the variance in the time-tradeoff values and 14% of the variance of the standard-gamble utilities could be explained by the best combination of RAND dimensions. These results suggest that answers to descriptive health-status questions cannot reliably predict standard-gamble utilities or time-tradeoff values. Key words: quality of life; health status; utility assess ment ; peripheral vascular diseases; intermittent claudication. (Med Decis Making 1996;16:217-225)

The authors assessed the relationship between the standard-gamble utility measure and the RAND-36 health-status dimensions, taking into account possible heterogeneity among patients in the weights they assign to different health-status dimensions. A questionnaire including both measures was completed by 68 patients with symptomatic peripheral arterial disease. Conventional multiple regression analysis, assuming a ho mogeneous relationship for the total population between the standard-gamble utility and the RAND-36 health-status dimensions, demonstrated that only the dimension social functioning was significant (p < 0.05), which accounted for 10% of the variation. Assuming that the population consisted of two separate classes demonstrated superior representation of the data. Latent class analysis was used to estimate the unknown parameters and class memberships. In the first class, consisting of 65% of the patients, the relationship between the standard-gamble utility and the dimension general health perception was significant. The within-R² was 12%. The second class represented 35% of the patients and showed significant coefficients for the dimensions social functioning and role limitations due to physical problems, which accounted for 80% of the variation. The overall percentage of variation explained by latent class analysis was 49%. The results suggest that patients with symptomatic peripheral arterial disease belong to a variety of classes, all with class-specific relationships between the standard-gamble utility and the RAND-36 health-status dimensions. Key words: health status measure; health utility measure; latent class analysis; heterogeneity. (Med Decis Making 1996; 16:226-233)

Background. The system to allocate scarce transplantable livers has been criticized for not giving enough weight to the prognoses of the patients receiving the transplants, but little research has been done looking at how the public weights the relative impor tances of efficacy and equity in distributing the organs. Methods. This study was an experimental survey of prospective jurors asked to distribute transplantable livers among transplant candidates grouped according to their prognoses. The relative prog noses of the transplant candidates were varied across survey versions. Results. As the prognostic difference between transplant groups increased, the subjects became less likely to distribute the organs equally between them (p < 0.005). However, the subjects' willingness to base allocation on prognosis was moderated by a number of factors, including their understanding of how to use prognostic information and their attitudes toward using prognostic information for individuals versus groups. Thus, even when the relative prognoses of transplant groups differed by 60%, less than a fourth of the subjects were willing to give all the organs to the better-prognosis group. Con clusion. Many subjects feel that prognosis is an important consideration in allocating scarce livers. However, few are willing to base allocation purely on maximizing survival. Policies that base allocations purely on outcomes will violate the values of a significant portion of the public. Key words: transplantation; equity; prognosis; ethics; health pol icy. (Med Decis Making 1996;16:234-241)

One of the most serious sources of potential bias when using the contingent valuation (CV) method to assess willingness to pay (WTP) is implied-value cues, i.e., different types of starting-point bias. The possible existence of starting-point bias is serious, since it may be interpreted to mean that the responders' preferences are very unstable. While the empirical evidence from environmental economics on starting-point bias is mixed, an earlier study in health economics did not find any clear evidence of starting- point bias. However, in the study presented here, a clear presence of starting-point bias was found. In a Swedish survey of how and when patients take antisecretory drugs, the patients were asked about their willingness to pay for a medication that can be taken in relation to meals compared with one that must be taken at least one hour before meals and has the additional disadvantage that it interacts with contraceptive pills. Among the 105 respondents, 82 were willing to pay a sum in addition to the normal patient fee in order to obtain the drug that could be taken during meals. The 82 patients thereafter participated in a bidding game that could start at a low bid (SEK 20) or a high bid (SEK 1,000). On average, the patients were willing to pay an addi tional SEK 138 (1 SEK = 0.13 U.S. dollar, April 1995) to obtain the superior drug. However, the average WTP among the 42 patients who started at the low bid was 70 SEK, which should be compared to an average of 289 SEK among the 40 patients who initially were offered the high bid. Key words: contingent valuation; willingness to pay; starting-point bias. (Med Decis Making 1996;16:242-247)

To compare the costs of health care programs with the benefits, the values of changes in health status must be expressed in monetary terms. The development of methods to estimate willingness to pay for changes in health status is therefore of interest. This paper reports the results of a contingent valuation study measuring willingness to pay for reductions in angina pectoris attacks. An innovative study design allowed analysis of the data on willingness to pay using two approaches, a binary question and a bid ding-game technique. Percentage reductions in anginal attacks were varied randomly in different subsamples, and data were collected about angina pectoris status, attack rate, and income to test the internal validity of the contingent valuation method. Will ingness to pay for a 50% reduction in the attack rate for three months was estimated to be about SEK 2,500 ($345) with the binary approach, and about SEK 2,100 ($290) using the bidding-game technique. Regression analyses showed that income, angina pectoris status, attack rate, and percentage reduction in attack rate were all related to willingness to pay, in agreement with the authors' hypothesis. Key words: willingness to pay; contingent valuation; angina pectoris; cost-benefit analysis. (Med Decis Mak ing 1996;16:248-253)

The Quality of Well-Being Scale (QWB) quantifies health-related quality of life with a single number that represents community-based preferences for combinations of symptom/problem complexes, mobility, physical activity, and social activity. The aim of this study was to compare preferences of a long-term care population with those of the general population, determine whether preferences vary by the age of the hypo thetical (target) person depicted in the health-state case description, and derive weights for new symptom/problem complexes of particular relevance to frail, older individuals. A sample of 38 female and 12 male long-term care residents with an average age of 86 years was asked to rate health-state scenarios that combined the four health domains of the QWB. This sample rated quality of life 0.10 units lower on average (on a 0-1 scale) than did the general population sample from which the QWB preferences were originally developed. Ratings of the same health state for younger versus older target persons did not differ significantly (all p values > 0.05 for t statistics). Weights derived for 11 new symptom/problem complexes were: disturbed sleep (-0.252), sit-to-stand requires maximal effort (-0.259), lonely (-0.265), walking a short distance causes extreme fatigue (-0.273), agitated (-0.284), hallucinating (-0.355), incontinent (-0.359), unable to control one's behavior (-0.360), urinary catheter (-0.374), restrained in bed or chair (-0.374), and feeding tube through the nose or stomach (-0.402). These new weights increase the relevance of the QWB for cost-utility evaluations of health interventions for long-term care residents. Key words: health state preferences; long-term care; patient preferences; Quality of Well-Being Scale; quality of life; age factors. (Med Decis Making 1996;16:254-261)

The probability-tradeoff technique may be used to assess treatment preferences in dichotomous choices. In this feasibility study, it was used to elicit benign prostatic hyperplasia (BPH) patients' attitudes towards three different treatments. Eighty-seven male outpatients used rating scales and the standard gamble to indicate the extents to which they were free of BPH symptoms. Paired descriptions of "watchful waiting" (WW), treatment with an alpha blocker (AB), and transurethral resection of the prostate (TURP) were presented, and the probability-tradeoff technique was used to obtain treatment-preference scores. The tradeoff task identified six internally consistent pref erence-order subgroups. The majority (n = 55; 63.2%) were in the two subgroups in which TURP was the least-preferred treatment. Compared with the other respondents, the members of these two subgroups reported significantly higher utilities for their BPH symptom status (89 vs 79; t = 2.87; p < 0.0005). Within each subgroup, preference scores for the middle- and top-ranked treatments were computed relative to the bottom- ranked treatment; for both WW and AB, significant across-subgroup differences were observed. In this preliminary study the probability-tradeoff technique was feasible, able to identify unique preference-order subgroups, and able to generate apparently mean ingful preference scores in a clinical situation involving three alternative treatments. Further development of tradeoff tasks as the value-clarification component of decision aids for individual patients seems warranted. Key words: patients' preferences; deci sion aids; benign prostatic hyperplasia; utilities; probabilities; risk aversion. (Med Decis Making 1996;16:262-272)

Mind and environment evolve in tandem—almost a platitude. Much of judgment and decision making research, however, has compared cognition to standard statistical models, rather than to how well it is adapted to its environment. The author argues two points. First, cognitive algorithms are tuned to certain information formats, most likely to those that humans have encountered during their evolutionary history. In par ticular, Bayesian computations are simpler when the information is in a frequency format than when it is in a probability format. The author investigates whether fre quency formats can make physicians reason more often the Bayesian way. Second, cognitive algorithms need to operate under constraints of limited time, knowledge, and computational power, and they need to exploit the structures of their environments. The author describes a fast and frugal algorithm, Take The Best, that violates standard principles of rational inference but can be as accurate as sophisticated "optimal" mod els for diagnostic inference. Key words: Bayes' theorem; bounded rationality; infor mation format; probabilistic reasoning; satisficing; training; medical education. (Med Decis Making 1996;16:273-280)

The author's purpose is to urge the constructive convergence of two current judgment and decision-making research paradigms. He shows why the heuristics-and-biases approach and the lens-model approach should be placed in the context of two very different metatheories, the coherence metatheory and the correspondence metatheory. The differences between the two research paradigms thus become apparent; they speak to different problems and appeal to different criteria for evaluating performance. Bringing the two into a constructive relationship to one another, however, will not only double the store of knowledge regarding diagnostic judgment and decision making, but also enhance efforts to achieve a cumulative discipline. Isolating these research paradigms from one another—as is done now—stifles theoretical generality, frag ments knowledge, and confuses medical decision makers. An example of how con vergence can be achieved is provided. Key words: diagnostic judgment; Bayes' the orem ; lens model. (Med Decis Making 1996;16:281-287)

There is growing interest in the application of cost-benefit analysis (CBA) as a tech nique for the economic evaluation of health care programs. A distinguishing feature of CBA is that costs and benefits are expressed in the same units of value—typically money. A popular method for estimating money values for health care programs is the use of willingness-to-pay (or accept) survey techniques known as contingent valuation. This paper presents a conceptual framework to help in the interpretation or design of contingent valuation studies in health care. To be consistent with the theory upon which CBA is built, the authors consider what types of questions should be asked of what populations. They conclude that studies undertaking contingent valuation should dis tinguish between compensating variation and equivalent variation, and recognize that respondents can be gainers or losers in utility and therefore should be asked willing ness-to-pay (or accept) questions as appropriate. Current critical-appraisal guidance in the health care literature for CBA is poor and unlikely to offer useful demarcation between good and bad CBA studies. More work is needed exploring whether recently issued guidelines for contingent valuation in environmental damage assessment are applicable to health care studies. Key words: cost-benefit analysis; contingent valu ation ; willingness to pay. (Med Decis Making 1996;16:288-299)