Objective: To perform a targeted systematic review of evidence regarding the benefits and harms of screening for asymptomatic carotid artery stenosis in the general population to support the update of the USPSTF’s 2014 D recommendation for this topic.
Data Sources: We conducted a literature search of MEDLINE, PubMed Publisher-Supplied Records, and the Cochrane Central Register of Controlled Trials (CENTRAL) from January 1, 2014, to February 14, 2020. In addition, we conducted ongoing surveillance of relevant literature through November 20, 2020.
Study Selection: We screened 2,374 abstracts and 144 full-text articles against a priori inclusion criteria. Retrospective analyses of vascular surgical registries were limited to data collected in the United States.
Data Analysis: Working independently, two investigators critically appraised each article that met inclusion criteria using design-specific criteria. We abstracted and narratively synthesized data from included studies. The results discussed in this report are limited to studies published since the previous review to support the 2014 recommendation.
Results: No eligible studies were identified that directly examined the benefits or harms of screening for asymptomatic carotid artery stenosis. Since the last USPSTF recommendation on this topic, two limited, fair-quality, prematurely terminated trials reported mixed results for the comparative effectiveness of carotid revascularization (carotid endarterectomy [CEA] or carotid artery stenting [CAS]) plus best medical treatment (BMT) compared with BMT alone. The SPACE-2 trial (N=316 reported no difference in composite outcome of stroke or death (30 days) or ipsilateral ischemic stroke (1 year) after CEA (unadjusted hazard ratio [HR] 2.82 [95% CI, 0.33 to 24.07]) or CAS (unadjusted HR 3.50 [95% CI, 0.42, 29.11]) compared with BMT in the 1-year interim publication. The smaller AMTEC trial (N=55) reported a statistically significantly lower composite risk of nonfatal ipsilateral stroke or death among the carotid endarterectomy (CEA) arm at 3.3 median years of followup (calculated unadjusted HR 0.20 [95% CI, 0.06 to 0.65]). Since the previous report, two fair-quality trials, two national datasets, and three surgical registries met our inclusion criteria reporting harms associated with CEA (N=1,903,761) or carotid artery stenting (CAS) (N=332,103). Overall, the rates of most postoperative adverse events were highest among analyses of national databases (Medicare data and National Inpatient Sample [NIS]), with lower rates reported in trials and surgical registries. Within the national databases and surgical registries, rates of 30-day postoperative stroke or death following CEA ranged from as low as 1.4 percent in the Vascular Quality Initiative (VQI) to as high as 3.5 percent in the Medicare database. Thirty-day postoperative mortality ranged from 0.5 percent in the Vascular Study Group of New England (VSGNE) to as high as 1.1 percent in the Medicare database for CEA. Thirty-day postoperative stroke rates following CEA ranged from 0.5 percent in the VSGNE to 1.5 percent in the American College of Surgeons National Surgical Quality Improvement Program (ACS NSQIP). For the CAS procedure, 30-day stroke or death ranged from 2.6 percent in the VQI to 5.1 percent in Medicare. Thirty-day postoperative mortality after CAS ranged from 1.1 percent in the VQI to 3.1 percent in the Medicare database. Thirty-day postoperative stroke rates following CAS were only reported in the VQI at 1.8 percent. Rates of postoperative harms within the trials were generally underpowered to detect outcomes such as postoperative mortality. Within the SPACE-2 trial, the composite of 30-day postoperative stroke or death was reported at 2.5 percent following both CAS and CEA. Perioperative stroke was reported in one patient (3.2%) following CEA in the AMTEC trial. The other most common harms reported within trials included hematoma, facial nerve lesion, and contrast agent incompatibility.
Limitations: We identified no trials of screening versus no screening in unselected general populations or examining direct screening harms. There were few new trials, all with methodologic concerns, examining the important question of the comparative effectiveness and harms of revascularization plus best medical treatment compared with best medical treatment alone. Selection bias and measurement bias presented serious validity concerns for complication rates reported in the administrative databases and surgical registries. The procedural complication rates of patients categorized as “asymptomatic” in the harms studies may not be generalizable to the rates that may be expected in a population of screen-detected patients (who would be expected to have lower complication rates compared with populations with any neurologic symptoms or remote history of TIA or stroke) or procedures performed outside of trials by less-selected operators (who may be expected to have higher complication rates compared with highly selected operators at high volume centers).
Conclusions: There are no population-based screening trials addressing the benefits and harms of screening for carotid artery stenosis. Limited new evidence has emerged to determine the benefits of carotid revascularization over contemporary best medical management in asymptomatic patients. The ongoing CREST-2 and ECST-2 trials will be the largest trials to address this issue. Large national administrative databases and surgery registries suggest that postoperative 30-day stroke/death rates vary widely—1.4 to 3.5 percent for CEA and 2.6 to 5.1 percent for CAS—suggesting that there may be a wide variation in complication rates likely attributable to patient and operator selection.