Introduction: Based largely on the inclusion criteria of clinical trials, the degree of cervical carotid artery stenosis is often used as an indication for stent placement in the setting of carotid atherosclerotic disease. However, the rigor and consistency with which such stenosis is measured outside of clinical trials is unclear. We compared the percent stenosis as measured by real world physician operators to that measured by independent expert reviewers. Hypothesis: We hypothesized that the documented degree of stenosis, termed operator-reported stenosis (ORS), from real world facilities performing carotid stent placement would be larger than the reviewer-measured stenosis (RMS) as assessed by clinicians experienced in treating carotid artery disease. Methods: Images were selected from patient cases used for carotid stenting facility accreditation. Data collected included demographics, National Institutes of Health Stroke Scale, modified Rankin Score, and the documented degree of stenosis (ORS). The ORS was compared to the RMS, derived from a panel of expert clinicians. Results: A total of 68 angiograms were reviewed from 39 symptomatic and 29 asymptomatic patients. The RMS values demonstrated excellent agreement with an intra-class correlation of 0.80. The median ORS was 90.0% (IQR 80.0, 90.0) versus a median RMS of 61.1% (IQR 49.8, 73.6), with a median difference of 21.8% (IQR 13.7, 34.4), p < 0.001. This discrepancy persisted within several subpopulations and was larger among asymptomatic patients and those treated at facilities granted delayed accreditation. Based on RMS values, 72% of symptomatic patients and only 10% of asymptomatic patients would meet the Centers for Medicare and Medicaid Services criteria for carotid stent placement. Conclusions: Real world operators tend to overestimate cervical carotid artery stenosis compared to external expert reviewers. Since decisions regarding carotid revascularization are often based at least partially on percent stenosis, such measuring discrepancies inevitably lead to a higher volume of procedures, which may in turn represent a degree of procedural over utilization.
Read full abstract