Skip to main content icon/video/no-internet

Scaling is the process of numerically measuring a health state utility. Utility is a global, composite, preference-based measure of health-related quality of life. Utility-based measures ask respondents to indicate their preference or desire for a health state, either their own or a hypothetical description. Utilities are scaled from 0 (death) to 1 (full or perfect health), although negative values can be assigned to health states considered to be worse than death. Utility is particularly valuable as a quality weight for length of life. In many decision analyses, cost-effectiveness analyses, and clinical studies, the main outcome is quality-adjusted life years, or QALYs, which are calculated by multiplying length of life by utility. Thus, 10 years in perfect health equal 10 QALYs, while 10 years in a health state with a utility of .75 equal 7.5 QALYs.

There are several standard scaling methods to obtain health state utilities. The most frequently used are the standard gamble, time trade-off (TTO), and rating scale. They differ in theoretical background, methodology, and outcome.

Theoretical Perspective

In the 19th century, utilitarian philosophers defined utility as the pleasure, good, or happiness, or prevention of pain, evil, or unhappiness produced by an object. Economists subsequently adapted utility to mean the satisfaction or pleasure that a consumer derived from a commodity or service. In both usages, utility was considered to be subjective, summable across individuals, and a motive for behavior.

In 1944, a mathematician and an economist, John von Neumann and Oskar Morgenstern, respectively, proposed that to the extent that utilities were preferences (i.e., an individual can say which object he or she prefers over another), utility could be numerically measured. Their method of measuring utilities involved making choices between alternative outcomes, where one included a risk. The value of any outcome could be inferred from how much risk an individual would take to avoid it. The axioms of von Neumann and Morgenstern defined how a rational individual ought to make decisions under conditions of uncertainty, that is, when decisions involved risk or chance, and provided proof of the existence of numerical utilities. These axioms became the foundation of expected utility theory, according to which an individual will behave or make choices to maximize his utility.

Scale Properties of Utilities

To be used as QALY weights, utilities must be measured on an interval scale. An interval scale is one in which changes of the same size have the same meaning anywhere on the scale, but 0 is an arbitrary value (such as temperature). Thus, a change in utility from .2 to .4 must be the same as a change from .7 to .9. The usual end points are death (0) and full health (1), but sometimes worst health is anchored at 0. An interval scale is a type of cardinal scale and allows all parametric statistical calculations.

Direct Scaling Methods for Measuring Utilities

Standard Gamble

The standard gamble (SG) offers respondents a series of choices between the certainty of spending a specified time period in the health state of interest and taking a hypothetical treatment that has an X% chance of immediate death and a (100 − X)% chance of full health. The health state of interest can be a patient's own health or a description of a hypothetical but plausible health state, often prototypical of a disease or condition. The chances of full health and death are varied, either by direct titration or by “ping-pong.” In the titration procedure, the first choice offered is between the health state of interest and a treatment that gives a 100% chance of full health and 0% chance of death. Once this is accepted, the chance of full health is decreased, usually by 5% at a time (95, 90, 85, etc.), and the chance of death increased (5, 10, 15, etc.), until the respondent indicates that he will not accept the gamble or cannot decide between the two choices. The point of indifference is either at this indecision or midway between the chance of full health that is accepted and the chance that is not accepted. In the ping-pong approach, the chance of full health is varied from high to low: 100%, 5%, 95%, 10%, 90%, and so on. The chance of death is always 100 minus the chance of full health. The ping-pong approach gradually closes in on the respondent's point of indifference between the choice of the gamble and the health state of interest. Utility for the health state is defined as 1 minus the probability of death at the point of indifference between the certainty and the risk. Thus, indifference between staying in current health and a potentially curative treatment with a 20% chance of death yields a utility of .80 for current health. If a health state is very undesirable the respondent should be willing to take a high risk of death to avoid it, and the respondent's utility for that health state will be low. Utilities for health states worse than death can be elicited by asking respondents to make a choice between certain death and a gamble in which full health occurs with a probability X and staying in the health state occurs with a probability of 1 − X. If the health state is very undesirable, the person would not take the gamble unless X is very high. Utility is calculated as −X/(1 − X) and therefore has a much larger range than the 0 to 1.0 limit for utilities for health states preferred to death. This can be corrected by assigning −1 to the least preferred health state and scaling the others between it and 0. Another method is to divide the negative utility by 1 minus itself; for example, if X = .95, utility = −19, and −19/(1 − (−19)) = −.95. The resulting values should be interpreted with caution.

...

  • Loading...
locked icon

Sign in to access this content

Get a 30 day FREE TRIAL

  • Watch videos from a variety of sources bringing classroom topics to life
  • Read modern, diverse business cases
  • Explore hundreds of books and reference titles

Sage Recommends

We found other relevant content for you on other Sage platforms.

Loading