HFSA Guideline Approach to Strength of Recommendation

Determining Strength. Although level of evidence is important, the strength given to specific recommendations is critical. The process used to determine the strength of individual recommendations is complex. The goal of guideline development is to achieve the best recommendations for evaluation and management, considering not only efficacy, but the cost, convenience, side effect profile, and safety of various therapeutic approaches. The HFSA guideline committee often determined the strength of a recommendation by the "totality of evidence," which is a synthesis of all types of available data, pro and con, about a particular therapeutic option.

Totality of Evidence. Totality of evidence includes not only results of clinical trials, but also expert opinion and findings from epidemiologic and basic science studies. Agreement among various types of evidence, especially from different methodologies, increases the likelihood that a particular therapy is valuable. Although many equate evidence-based medicine with the results of a few individual clinical trials, the best judgment seems to be derived from a careful analysis of all available trial data combined with integration of results from the basic laboratory and the findings of epidemiologic studies.

Scale of Strength. The HFSA guideline employs the categorization for strength of recommendation outlined in Table 1.3. There are several degrees of favorable recommendations and a single category for therapies felt to be not effective. The phrase "is recommended" should be taken to mean that the recommended therapy or management process should be followed as often as possible in individual patients. Exceptions are carefully delineated. "Should be considered" means that a majority of patients should receive the intervention, with some discretion involving individual patients. "May be considered" means that individualization of therapy is indicated (Table 1.3). When the available evidence is considered to be insufficient or too premature, or consensus fails, issues are labeled unresolved and included as appropriate at the end of the relevant section.