Summary of findings 1. Summary of findings table: Duplex ultrasound for diagnosing symptomatic carotid stenosis in the extracranial segments.

Review question:	What is the diagnostic accuracy of duplex ultrasound for detecting symptomatic carotid stenosis?
Population	Symptomatic patients (sudden visual loss, hemispheric TIA, and ischemic stroke) with suspected carotid artery stenosis
Target condition	Carotid artery stenosis
Index test	Duplex ultrasound
Reference Standard	DSA in 19 studies (Anzidei 2012; Borisch 2003; Chua 2007; Colquhoun 1992; Cui 2018; D'Onofrio 2006; Bray 1995; Eliasziw 1995; Faught 1994; Golledge 1999; Hammond 2008; Hansen 1996; Heijenbrok‐Kal 2006; Huston 1993; Knudsen 2002; Link 1997; Lubezky 1998; Nederkoorn 2002; Wolfle 2002); MRA in three (Borisch 2003; D'Onofrio 2006; Das 2009); CTA in four (Barlinn 2016; Belsky 2000; Das 2009; Lubezky 1998)
Importance	Diagnostic accuracy of DUS to identify carotid artery stenosis in symptomatic patients can improve the path in defining the best treatment option
Included studies	We included 22 studies, with a total of 4957 carotid arteries, mean sample size of 126 carotid arteries, ranging from 24 to 1011; the mean age of participants was 66.3 years (range 53 to 72 years), and the median proportion of men was 70% of included participants. Eighteen prospective studies, two retrospective and, in two studies, it was unclear whether there was a prospective or retrospective design
Risk of bias and applicability concerns	Risk of bias varied considerably across the included studies; we considered nine studies as being at high risk of bias and one as having unclear concern in the patient selection domain, mostly due to failure to include all people with a negative screen or poorly reported patient selection methods; four studies were judged as having a high risk of bias in the index test domain, mostly because of no prespecified thresholds; two as being at high risk of bias and seven at unclear risk of bias in the reference standard domain, as the studies were not blinded or blinding was not described; and the risk of bias in the flow and timing domain was high in 14 studies because not all patients were included in the analysis and it was unclear in another two. Applicability concerns were generally low; six studies were judged as having high concern on the patient selection domain mostly because of previous testing.
Limitations	Seventeen studies were judged as having high risk of bias in at least one domain. The use of velocity criteria with prespecified thresholds and time we accepted between the index test and reference standard (four weeks) led to a lot of studies' exclusions. There were also a lack of data on some carotid stenosis categories and reference standards.
Reference Standard	Studies	Carotid arteries	Summary sensitivity (95% confidence interval)	Summary specificity (95% confidence interval)	Consequences in a cohort of 1000
					Prevalence of the range of stenosis (median) *	Implications *	Quality and Comments
DSA
< 50%	4	1495	0.63 (0.48 to 0.76)	0.99 (0.96 to 0.99)	0.46	460 out of 1000 patients will have < 50% carotid artery stenosis. Of these, 291 (63%) would be correctly diagnosed and receive appropriate clinical treatment and 169 (27%) would receive unnecessary further investigation with another imaging method. Other 532 patients would receive appropriate further investigation, and eight would have no other tests performed and miss a chance for the right diagnosis and the possibility of carotid revascularization .	Limited number of studies Risk of bias: High or unclear in most domains
50‐99%	5	1536	0.97 (0.95 to 0.98)	0.70 (0.67 to 0.73)	0.51	510 out of 1000 patients will have 50‐99% carotid artery stenosis. Of these, 495 (97%) would receive appropriate further investigation with another imaging method, and 15 (3%) would not have any other tests performed and would miss a chance to receive the right diagnosis and the possibility of carotid revascularization. Overall, 147 would receive unnecessary further investigation with another imaging method, and 343 would receive no further investigation and appropriate clinical treatment	Limited number of studies Risk of bias: High or unclear in most domains
50‐69%	1	313	0.28 (0.17 to 0.41)	0.90 (0.85 to 0.93)	0.19	Meta‐analyses not conducted
70‐99%	9	2770	0.85 (0.77 to 0.91)	0.99 (0.96 to 0.99)	0.45	451 out of 1000 patients will have 70‐99% carotid artery stenosis. Of these, 383 (85%) would receive appropriate carotid artery revascularization and 68 (15%) would miss or delay the chance to carotid revascularization. Another 8 would receive inappropriate carotid artery revascularization and 542 would receive appropriate clinical treatment.	Limited number of studies Risk of bias: Low risk in all domains in 2 studies
Occluded	7	1212	0.91 (0.81 to 0.97)	0.95 (0.99 to 0.76)	0.18	180 out of 1000 patients will have carotid artery occlusion. Of these, 164 (91%) would receive appropriate clinical treatment. Another 41 would be false‐positive diagnosed with carotid occlusion and not have other tests performed, and miss a chance of the correct diagnosis and carotid revascularization. Other consequences would depend on the range of stenosis.	Limited number of studies Risk of bias: Low risk in all domains in 1 study Two studies only included patients with occlusion on DUS. Sensitivity analyses excluding them had impact on the results of specificity: 0.98 (95% CI: 0.97 to 0.99).
CTA
70‐99%	2	685	Range: 0.57 to 0.94	0.87 to 0.98	0.18	Meta‐analyses not conducted
Occluded	3	833	0.95 (0.80 to 0.99)	0.91 (0.99 to 0.09)	0.60	600 out of 1000 patients will have carotid artery occlusion. Of these, 570 (95%) would receive appropriate clinical treatment. Another 41 would be false‐positive diagnosed with carotid occlusion and not have other tests performed, and miss a chance of the correct diagnosis and carotid revascularization. Other consequences would depend on the range of stenosis	Limited number of studies Risk of bias: High or unclear in most domains 1 study only included patients with occlusion on DUS
MRA
50‐99%	1	31	0.88 (0.70 to 0.98)	0.60 (0.15 to 0.95)	0.84	Meta‐analyses not conducted
70‐99%	2	102	Range: 0.54 to 0.99	Range: 0.89 to 0.78	0.61	Meta‐analyses not conducted
CI: confidence interval; CTA: computed tomography angiography; DSA: digital subtraction angiography; DUS: duplex ultrasound; MRA: magnetic resonance angiography; TIA: transient Ischemic attack
* We calculated prevalence from the included studies by the reference standard. The prevalence values used to illustrate the review findings as absolute frequencies are the median from the included studies.
CAUTION: The results on this table should not be interpreted in isolation from the results of the individual included studies contributing to each summary test accuracy measure. These are reported in the main body of the text of the review.