Threshold Technique

The Threshold Technique measures individuals' attitudes toward a single key attribute of a particular “target” therapy. The procedure involves repeatedly varying the levels of the key attribute of primary interest and, with each variation, asking the respondent to choose between the target therapy and an alternate reference therapy.

The key attribute that is varied depends on the study purpose. For example, the attribute could be chance (e.g., the probabilities of side effects or benefits), time (e.g., time waiting for therapy or life expectancy), or distance to travel (e.g., for access to care). Compared with other methods for evaluating therapies, it is close to the Willingness-to-Pay method, in which cost is the key attribute that is varied.

The technique can be appropriately adapted to a wide range of clinical dilemmas in medical decision making. For example, one version can reveal patients' minimal required reduction in the long-term risk of a heart attack that they would want from a cholesterol-lowering agent (the target) before considering it worthwhile compared with lifestyle management alone (the reference). In another example, consider the choice between a current analgesic (the reference) and a new medication (the target) that offers potentially greater pain relief but carries a higher risk of gastric bleed. Different versions of the technique could be designed to reveal either the minimal amount of pain relief that patients would require or the maximal level of risk for gastric bleed that patients would accept before considering the new drug to be preferable to their current medication.

Therefore, the technique is very flexible. However, because it focuses on a single key attribute, it can appropriately address only narrowly defined and context-dependent research questions. Conceptually and procedurally, the technique is very different from other methods for evaluating therapies— including formal Decision Analysis, Conjoint Analysis, the Analytic Hierarchy Process, the Balance Technique, and the Leaning Scale, which work with multiple attributes in a more holistic manner and are designed for different research purposes.