Use of GRADE for assessment of evidence about prognostic factors: rating certainty in identification of groups of patients with different absolute risks.

The Grading of Recommendations, Assessment, Development and Evaluation (GRADE) approach to rating certainty in the results of research studies was initially developed for therapeutic questions. The approach considers: study design, risk of bias; inconsistency; imprecision; indirectness, publication bias; magnitude of effect; and dose-response. Questions about prognosis bear great relevance for decision-making in health care. Studies of prognosis can inform individuals about their likely outcome: for instance, in patients with a new diagnosis of cancer, are they likely to be alive or dead in five years. Further, prognostic studies can aid decisions in those considering treatment: for instance, is one's risk high enough to use medication to prevent cardiovascular disease. It follows that health care professionals and patients need to know how confident they can be regarding such prognostic estimates. We have previously provided guidance for using the GRADE approach to determine certainty in estimates of future events in broad categories of patients (overall prognosis). Prognostic studies may also provide more or less robust estimates of the association between patient characteristics (such as their age, sex, and coexisting illness) and undesirable or desirable outcomes. GRADE's approach to certainty of the evidence aims to inform clinicians and patients of the trustworthiness of the estimates from systematic reviews of studies addressing such individual prognostic factors