Preferences And Utilities

Modern Ayurveda

Ayurveda the Science of Life

Get Instant Access

When the outcomes are known in advance, patients can express a preference for one treatment or another. For example, if it can be stated that treatment will result in cure but at the cost of specified QoL disadvantages, patients can decide whether or not to accept the treatment. However, in most clinical situations there will be uncertainty regarding both the cure and the QoL outcomes. Usually we can state only that there is a certain probability of cure, and that this may be gained at possible QoL disadvantages. When preferences are assessed in the face of uncertainty they are called utilities, because a patient might make one selection under uncertainty but might express a different preference if it were known what outcome would ensue.


The simplest form of establishing preference ratings is by means of a visual analogue rating scale, in which the extreme anchors are usually "best possible QoL" and "worst possible QoL", or some equivalent wording. The patient is then asked to indicate on the 10-cm line the position of their current state, and also to mark positions corresponding to various scenarios such as their likely condition during or following therapy. For the least favourable state, "death" is often avoided because some patients may declare particular states of health to be worse than death; thus for alternative wording one might consider "worst imaginable state of health".

This method has been found to be efficient and easy to use, and appears to provide meaningful values for relative preferences of various states of health and treatment. It can be extended to include the concept of uncertainty, thereby providing utilities, if the scenarios include suitably phrased descriptions that indicate a risk of side-effects or disease progression; in this case, extra care is needed in choosing appropriately worded endpoints for the scale.


Time trade-off involves comparing QoL against length of survival. A typical strategy for evaluating TTO is to present a scenario under which health is impaired by specific disabilities or symptoms, and to ask the patient whether he or she would choose one year in perfect health or one year with impaired health; presumably, the healthy year would be selected. Then the duration of the healthy period is gradually reduced: "Would you choose 11 months in perfect health, or one year with impaired health?", and so on. At some stage equilibrium should be reached, and it may then be concluded that the value of impaired life is equivalent to a certain percentage of time relative to healthy life.

As we shall see, TTO is conceptually equivalent to the QALY approach (see Section 12.5), and might therefore seem attractive. Since it does not involve uncertainty, it is a method for eliciting patient preferences. However, many patients find the concepts difficult to apply.


The standard gamble method involves decisions in the face of uncertainty, where the uncertainty involves a risk of death or some other outcome. Thus SG attempts to estimate patient utilities.

For example, SG might be used to establish the value of anti-hypertensive therapy by offering the following alternatives to patients: "Suppose there is a P% chance of death within the following year if you do not take anti-hypertensive therapy, but on the other hand you would have to take therapy for the rest of your life and it has these side-effects. . . ." By varying the percentage, P. the point of indifference can be established. The value (1 - P) then provides the utility value for impairment due to this form of therapy.

As for TTO, many patients find the concepts of SG unrealistic and have difficulty in making consistent responses. One particular problem is that it is frequently difficult to provide realistic scenarios for some of the medical conditions and therapies that are under consideration.

Example from the literature

De Haes and Stiggelbout (1996) compared VAS, TTO and SG methods in 30 testicular cancer patients. In line with other reports, the VAS method yielded the lowest scores, and TTO was slightly lower than SG. The authors noted that since many patients are reluctant to trade survival for QoL, and are willing to accept high levels of toxicity for a relatively modest increase in survival time, it is perhaps not unexpected that TTO results in higher scores than VAS. Similarly, SG patients are asked to consider the possibility of immediate death, which is even less acceptable to many. De Haes and Stiggelbout suggest that the choice between the three methods might be made according to the disease and the intended application of the ratings; for example, in a surgical trial that involves a risk of early death, the SG approach might be preferred.


Whereas SG involves a gamble and the element of risk when comparing the value of different health states, and TTO uses varying time periods, willingness to pay introduces the concept of monetary value. The foundation for this is that people are accustomed to making decisions about how much they are willing to spend upon most things relating to lifeā€”from small items such as food and clothing, and medium-cost decisions such as annual holidays, through to major expenses including car and house. The amount that an individual is willing to pay is an indicator of the utility or satisfaction that they expect to gain from the particular commodity.

Various methods have been used to elicit WTP values, including basic questions such as "What is the most that you would be willing to pay for . . .". A variation on this is to present a list of options or a set of cards containing "bids" of increasing amounts, from which the respondent selects the amount they would be willing to pay.

WTP has rarely been used in QoL research. CONJOINT ANALYSIS (CA)

A newer technique for assessing preferences of health states is conjoint analysis, which has been used extensively in market research and transport economics. CA involves constructing a number of realistic scenarios that represent combinations of different health states (item levels). Obviously there are too many possible states to be able to present them all (see Section 12.3), and so particular scenarios must be selected. The patient is then presented with two or more of the scenarios, and asked to rank them, rate them or indicate their preferred option. Since the majority of people are accustomed to making what are effectively pairwise comparisons and decisions on a daily basis, Ryan (1999) has argued that this may be the preferred approach for CA.

The resultant data consist of binary preferences, and must therefore be analysed using logistic (or the related probit) regression models; some of the models are complex, requiring specialised software for their fitting (Ryan, 1999).

Was this article helpful?

0 0
Fitness Wellness For You

Fitness Wellness For You

Achieve the Fitness and Wellness for You that you have always wanted by learning the facts so you can take the right steps to maximize your health. Learn How to Achieve Real Fitness and Wellness for a Healthy Body, Mind and Spirit to Improve Your Quality of Life in Today's World. Receive Valuable Information to Discover What Really Matters and What Actually Works in Finding Genuine Wholeness for All Aspects of Your Being.

Get My Free Ebook

Post a comment