Patient-Reported Outcomes

8. Interpretability

A final key property of an HRQL measure is Interpretability.

  • For discriminating between groups, we could ask whether a particular score signifies that a patient is functioning normally or has mild, moderate, or severe impairment.
  • For evaluating change over time, we might ask whether a particular change in a score rep­resents a trivial, small but important, moderate, or large improvement or deterioration.
  • A number of strategies are available to make PRO scores interpretable (Guyatt et al., 1991), such as:
    1. Classify patients into those who had impor­tant improvement as well as those who did not and examine the changes in score in the two groups; 
    2. Inter­pret observed changes in PRO measures in terms of elements of those measures that will be familiar to read­ers (for instance, descriptions of changes in mobility); or
    3. Determine how scores in PRO measures relate to marker states that are familiar and meaningful to clini­cians.

Data suggest that small, medium, and large ef­fects correspond to changes of approximately 0.5, 1.0, and greater than 1.0 per question for instruments that present response options on seven-point scales (Jaeschke et al., 1989).

Example 5

  • In a domain with 6 items, changes of 3 or 4 represent small effects, changes of 5 or 6 represent moderate effects, and changes of 7 or more represent large effects. Investigators used this information to in­terpret a recent trial that showed use of bronchodilators results in a small but clinically important improvement in dyspnea, fatigue, and emotional function in patients with chronic airflow limitation (Guyatt et al., 1987).
  • In a study (Thompson et al., 1988) of patients with arthritis, a change of 0.02 points in the Quality of Well-Being utility instrument was equivalent to all treated patients improving from moving their own wheelchair without help to walking with physical limi­tations. The availability of data to improve the Interpretability of PROs is likely to increase exponentially in the next decade.