GRADE guidelines: 6. Rating the quality of evidence-imprecision

摘要: GRADE suggests that examination of 95% confidence intervals (CIs) provides the optimal primary approach to decisions regarding imprecision. For practice guidelines, rating down quality evidence (i.e., in estimates effect) is required if clinical action would differ upper versus lower boundary CI represented truth. An exception this rule occurs when an effect large, and consideration CIs alone a robust effect, but total sample size not large number events small. Under these circumstances, one should consider for To inform decision, can calculate patients adequately powered individual trial (termed "optimal information size" [OIS]). continuous variables, we suggest similar process, initially considering limits CI, subsequently calculating OIS. Systematic reviews require somewhat different approach. If excludes relative risk (RR) 1.0, or exceeds OIS criterion, precision adequate. includes appreciable benefit harm (we RR under 0.75 over 1.25 as rough guide) imprecision may be appropriate even criteria are met.