Quality of Evidence in European Stroke Organisation and American Heart Association Stroke Guidelines

Evidence-based practice relies on clinical guidelines, whose recommendations depend on the quality, relevance, and validity of supporting research. We evaluated the class/strength and level of evidence (LOE) or quality of evidence (QOE) supporting American Heart Association/American Stroke Association and European Stroke Organisation guideline recommendations, and examined temporal changes in LOE.

Stroke guidelines from American Heart Association/American Stroke Association (1995-2025) and European Stroke Organisation (2014-2025) were identified through society websites and EMBASE/MEDLINE. Eligible documents contained recommendations with class/strength and LOE/QOE. Consensus statements were excluded. Since 2006, American Heart Association/American Stroke Association has classified LOE as A (multiple or large randomized-controlled trials), B (single trial or observational studies), or C (expert opinion). European Stroke Organisation applies the Grading of Recommendations Assessment, Development, and Evaluation system (high, moderate, low, and very low QOE).

Across 1102 recommendations in 9 current American Heart Association/American Stroke Association stroke guidelines, 156 (14.2%) were supported by LOE A, 559 (50.7%) by LOE B, and 387 (35.1%) by LOE C. Of 407 class I recommendations (ie, should do), and 117 class III recommendations (ie, should not do), 116 (22.1%), 258 (49.2%), and 150 (28.6%) were supported by LOE A, B, and C, respectively. Although the number of recommendations increased across guideline updates (median, 22 [interquartile range, 25th-75th percentiles, 18.0-42.0]), the proportion supported by LOE A declined (median, -4.6% [interquartile range, -7.8 to -0.8]). Across 260 recommendations in 30 European Stroke Organisation guidelines, 19 (7.3%) were supported by high, 62 (23.8%) by moderate, 81 (31.2%) by low, and 98 (37.7%) by very low QOE. Among 90 strong recommendations, 18 (20.0%) were supported by high QOE, and 66.7% of guideline topics had no recommendations supported by high QOE. There was insufficient evidence to make recommendations for 123 (32.7%) clinical questions.

Due to limited randomized data for many important clinical questions, most stroke guideline recommendations are based on low-to-moderate-quality evidence. These findings emphasize the need to improve the funding, design, and delivery of efficient, patient-focused clinical trials.