Table 2.

Confidence in diagnosis, proportion of correct responses and strength of agreement among 146 clinicians worldwide assigning diagnoses to ten real-life scenarios of acute neurological complications of COVID-19.

Neurological Syndrome⁎	Number of respondents (n)	Mean confidence in diagnosis (0−10)⁎⁎	Proportion of correct diagnosis (%)	95% CI (%)	Strength of agreement (Fleiss Kappa)	95% CI (Fleiss Kappa)
Cerebral venous thrombosis	120	8.77	95.8	92.3–99.4	0.676	0.670–0.682
Guillain-Barré syndrome	119	8.89	92.4	87.7–97.2	0.622	0.616–0.628
Headache	119	7.67	91.6	86.6–96.6	0.702	0.696–0.708
Myelitis	127	8.47	81.9	75.2–88.6	0.688	0.682–0.694
Seizure	123	7.34	80.5	73.5–87.5	0.599	0.593–0.605
Ischaemic Stroke	130	7.42	79.2	72.3–86.2	0.476	0.470–0.482
Cranial neuropathy	137	7.15	78.1	71.2–85.0	0.710	0.704–0.716
Encephalitis	125	7.32	72.8	65.0–80.6	0.422	0.416–0.428
Psychosis	119	7.14	53.8	44.8–62.7	0.409	0.403–0.415
Encephalopathy	146	6.80	43.2	35.1–51.2	0.257	0.251–0.263

^⁎Cases are ranked by the proportion of respondents assigning the correct diagnoses, such that the cases with the lowest numbers of correct answers are at the bottom. The overall agreement across all diagnoses was moderate (Fleiss Kappa = 0.443, 95% CI 0.441–0.445).

^⁎⁎Mean confidence in diagnosis was derived for each scenario by transforming ordinal variables from Likert scale into a continuous (0 to 10) scale, as described previously [19]. Score 10 would indicate that all respondents rated their confidence as “very high” whereas score 0 would indicate that all respondents rated their confidence as very low.