Box 2: Description of levels of evidence reported in Canadian Stroke Best Practice Recommendations33

Level of evidence	Criteria
A	Evidence from a meta-analysis of RCTs or consistent findings from 2 or more RCTs. Desirable effects clearly outweigh undesirable effects or vice versa. Phrases used in recommendations with this level of evidence include “strong recommendation”; actions “should be (or not be) done.”
B	Evidence from a single RCT or consistent findings from 2 or more well-designed nonrandomized or noncontrolled trials, and large observational studies. Meta-analysis of nonrandomized or observational studies. Desirable effects outweigh or are closely balanced with undesirable effects, or vice versa. Phrases used in recommendations with this level of evidence include “is recommended”; “should be considered”; and in some cases where there is strong agreement by the writing group, “actions should be (or not be) done on most or specific groups as applicable.”
C	Writing group consensus on topics supported by limited research evidence. Desirable effects outweigh or are closely balanced with undesirable effects or vice versa, as determined by writing group consensus. Phrases used in recommendations with this level of evidence include “may be considered” or “is reasonable.”
Clinical consideration	Reasonable practical advice provided by consensus of the writing group on specific clinical issues that are common or controversial and lack research evidence to guide practice. No evidence levels are assigned to clinical considerations.