Every guide to decision making emphasizes the importance of goals. Decision makers must clarify their goals when making a decision, lest they make choices that will not serve their ends. To consider alternatives without knowing one’s goals is to let the tail wag the dog.

Practically speaking, most medical decision models don’t (and perhaps can’t) consider patient goals directly. Good clinicians, however, must (and do).

Medical decision models typically adopt two goals that are both common and easily amenable to measurement:

1. Live as long as possible
2. Maintain as functional/good a state of health as possible

In fact, these two common goals form the basis of the QALY model for evaluating health outcomes (and thus guiding medical decisions), about which I’ll have more to say over time.

Focusing on the common goals makes sense for health policy decisions, and is a convenient simplification. Incorporting individual patient goals is hard: they differ from patient to patient, and usually only the patient herself has really good insight into her goals. But for good clinical decision making, incorporating individual goals is also necessary, because rational people are willing to sacrifice some of their life or health in order to achieve other important goals.

Key questions for the physician seeking to help a patient make significant medical decisions, then, are which goals are important to this patient, and how important are they relatively to one another?

I’ll discuss goal clarification — helping a patient determine what their key goals might be — in another post. Here I want to describe a method for determining the relative importance of a set of goals in the context of a particular medical decision.

The procedure I’ll be describing is a variation of swing weighting, and is fully described in the seminal decision analysis book Decision Analysis and Behavioral Research, by Detlov von Winterfeldt and Ward Edwards (sadly no longer in press, but available from many used bookdealers).

To use swing weighting with a set of goals and a set of treatment outcomes, you determine from the patient, how each treatment outcome would impact each of their goals.

For example, if a patient has three goals:

1. Live as long as possible
2. Continue working at my job
3. Keep my family financially secure

and is considering three treatments (A, B, and no treatment) with known outcomes, they might perceive the situation to look like this:

To determine the relative importance of their three goals, we ask which goal could result in the most important swing from worst possible achievement of that goal to best possible achievement. That is, if the patient were the worst possible scenario (no more than two years to live, unable to work for months, and sufficient insurance coverage), which one goal’s outcome would they want to swing to the best possible achievement? Would they prefer to improve their longevity (from “no more than 2 years” to “5+ years”), their ability to work (from “unable to work for months” to “immediately resume work”), or their financial security (from “sufficient coverage” to “sufficient coverage”)? If the patient prefers to swing Goal 1, that is considered her most important goal, and she can then proceed to choose which goal she would next choose to swing, out of those that remain.

The beauty of swing weighting is that it recognizes the range of outcomes on each goal is crucial in making judgments about their relative importance. Even if the family’s financial security is the patient’s most important goal in the abstract, its importance is low (actually nil) in the decision above, because none of the options will produce a change in that goal.