Skip to main content
Fig. 3 | BMC Medical Research Methodology

Fig. 3

From: Measuring clinical uncertainty and equipoise by applying the agreement study methodology to patient management decisions

Fig. 3

Thrombectomy decisions. Legend: Panel a shows the proportion (%) of decisions to perform thrombectomy (in percentages) for all raters and among each specialty. Black dots represent the individual results of each of the 86 clinicians. The bar graphs show similar proportions of decisions between neurologists and interventional neuroradiologists (INRs), but they hide individual discrepancies among physicians, shown here by black dots, revealing a wide range of decisions. Panel b shows, for each patient, the proportions (%) of thrombectomy decisions. This panel better illustrates the spectrum of results in various patients, as it shows that some cases had almost unanimous decisions for (complete/almost complete blue bar at the top) or against thrombectomy (complete/almost complete red bar at the bottom part). However, a significant proportion of cases (in the middle) reveal wide disagreements. None of these panels can give an overall idea of the degree of agreement in the study. Panel c shows the levels of agreements (through kappa values) in a bar graph. It shows that thrombectomy decisions lack reliability (i.e kappa value is below 0.6) for all raters and also within each subspecialty (vascular neurologists and interventional neuroradiologists)

Back to article page