Dr. Wells Replies

To the Editor: Dr. Bursztajn raises the question of whether the authors of treatment-outcome studies should justify their outcome selection within a broader scheme that includes effects relevant to practice and policy that are difficult to measure. The simple answer is “yes,” but this is not a simple question.

Practice and policy decisions involve tradeoffs among treatments and their expected outcomes under conditions of constrained resources, uncertainty, and personal distress. The information required to inform those decisions ideally includes all expected benefits and costs, including respect for individual autonomy and societal implications, of the alternative actions. In addition, one would desire a validated process to support the parties in making decisions that reflect their authentic values (from the 1990 book by Bursztajn et al.). Obtaining such data and using them well are the challenging tasks at the heart of both clinical practice and the science of practice. For example, there are known theoretical and technical problems in measuring preferences and reconciling individual and societal perspectives (1). Dr. Bursztajn refers specifically to an ethics outcome (i.e., autonomy) and an attribute of values (i.e., authenticity). These are among other ethics outcomes and values, such as distributive justice, that are seldom considered in treatment outcome studies. How can we develop a broad basis for decisions that include them?

First, the parties in health care and researchers must become more familiar with the range of outcomes of individual and social salience. Such a shared conceptual framework is difficult to accomplish even across two areas, much less several, including clinical, economic, and ethical outcomes. Second, interventions that use scientific data to inform practice must be developed and tested widely, and their implications for these diverse outcomes must be understood. While it is difficult to assess values, it may not be much more difficult than studying quality of life, which is not easy to measure or analyze. Yet smaller studies may not be able to study either type of outcome owing to precision problems. The field can accomplish more, however, by collaborating on fewer studies and studies that have scope and scale or coordinating meta-analyses. It is not clear, however, if diverse investigator groups can or will engage in such planning or if such consolidation of research resources compromises the discovery process.

Meanwhile, studies can achieve progress by improving measures of diverse outcomes and understanding their relationships and determinants. For example, we are assessing the effects of a quality improvement program for depression on clinical, economic, and selected ethical outcomes in the Partners in Care program (2). Tomorrow’s clinicians and researchers may need to consider these and broader effects routinely if improved health care information systems enable us to make more richly informed decisions in real time. I hope my recent article represents a step toward the preparation of clinicians and researchers for this agenda and that others, like Dr. Bursztajn, will stimulate discussion as to how best to achieve a useful database and an implementation process for practice decisions.