Bias in the use of randomized trials for carotid stenosis management.

backgroundCarotid artery procedures, such as surgery and stenting, although associated with significant risks and costs, are often recommended in guidelines which cite12- to 34-year-old randomized trial evidence of benefit; however, these recommendations exist although there is no evidence these procedures benefit patients who receive only current optimal medical treatment  (encouragement of a healthy lifestyle and appropriate use of medication).

objectiveTo examine whether bias exists in the use of randomized trial evidence and its impact on guideline recommendations. MATERIAL AND

methodsExamples of how bias underpins endorsement of carotid procedures for patients with asymptomatic or symptomatic carotid stenosis were sought from available literature. .

resultsMany forms of procedural bias were identified involving the need for randomized trials, and their design and interpretation. Fundamental problems included failure to first adequately measure outcomes with non-invasive treatment alone, lack of appreciation of quality non-randomized trial measurements of risk in determining need for randomized trials and their applicability in routine practice, poor randomized trial methods with biased comparisons, inaccurate definitions of target populations, confusion of efficacy and safety outcomes, too much reliance on statistical rather than clinical significance and biased use of terminology to make procedures sound more effective.

conclusionProcedural bias in design and interpretation of randomized trials has resulted in widespread loss of understanding of how to optimize outcomes in patients with carotid artery stenosis. Current guidelines reflect the cumulative impact of this bias and are an excellent starting point for efforts to improve prevention of stroke and other vascular disease complications; however, there is also need for clinicians, policy makers, health service funding bodies, educators and the general public to assist.