Development of outcome-based practice guidelines: a method for structuring problems and synthesizing evidence.

Abstract

BACKGROUND: The growth in guidance development projects has focused attention on the methods used in developing the guideline. For a guideline to be sound it should be linked on the basis of scientific evidence to the very health outcome that the guideline is designed to promote. METHODS: Structuring a health intervention as an influence diagram, a decision model (1) allows for the identification of the relevant benefits, harms, and costs that may result from an intervention; (2) provides an explicit link between the intervention and these outcomes, a crucial prerequisite for the development of an outcome-based guideline; and (3) identifies the evidence that must be synthesized to predict the effect of the intervention on the health outcomes. EXAMPLE: In the development of a guideline related to prevention of opportunistic infections in HIV-infected persons, we would define the interventions (for example, use of medication for PCP pneumonia), the intended health outcome (a potential reduction in the number of opportunistic infections), and the evidence that demonstrates that the intervention produces the desired outcome. If PCP prophylaxis is delayed, the HIV-infected person is exposed to a undue risk of PCP, with its attendant morbidity and mortality. If it is initiated too early, the person incurs excess monetary costs and may experience additional side-effect-associated morbidity. EXAMPLE: The intervention in question is screening for HIV infection, and the outcomes of interest are the medical benefits and harms associated with screening and the financial costs (and savings) that a screening program would incur. Screening for HIV infection differs from many clinical questions because it has potential benefit both to the persons screened and to public health if the screened person reduces risk behaviors that might transmit HIV infection. CONCLUSIONS: Structuring a problem with an influence diagram: delineates an explicit link between interventions and outcomes; focuses the questions to be addressed (a series of more sharply defined questions, each of which we may be able to answer based on direct evidence, replaces a much broader question [should we screen for HIV?], which cannot be answered directly); and highlights the importance of a clear, unambiguous statement whose benefit and costs are under consideration.