Rapid progression of carotid stenosis was rare in a large integrated healthcare system during an eight-year period

Abstract

Objective: Few studies have evaluated the rapid progression of carotid stenosis on a large scale. We created a custom software algorithm to analyze an electronic medical record database to examine the natural progression of carotid stenosis, identify a subset of patients with rapid progression, and evaluate the specific patient risk factors associated with this rapid progression.

Methods: Patients in a large integrated healthcare system who had undergone two or more carotid ultrasound scans from August 2010 to August 2018 were identified. We did not distinguish between those with an established carotid stenosis diagnosis and those with a screening ultrasound scan. We used our novel algorithm to extract data from their carotid ultrasound reports. The degrees of carotid stenosis were categorized as follows: level 1, 0% to 39%; level 2, 40% to 59%; level 3, 60% to 79%; level 4, 80% to 99%; and level 5, complete occlusion. The primary endpoint was rapid vs slow progression of carotid stenosis, with rapid progression defined as an increase of two or more levels within any 18-month period of the study, regardless of the date of the initial ultrasound scan. The association of the demographic and clinical characteristics with rapid progression was assessed by univariable and multivariable logistic regression.

Results: From a cohort of 4.4 million patients, we identified 4982 patients with two or more carotid ultrasound scans and a median follow-up period of 13.1 months (range, 0.1-93.7 months). Of the 4982 patients, 879 (17.6%) had shown progression of carotid stenosis. Only 116 patients (2.3%) had had progression to level 4 (80%-99% stenosis) from any starting level during a median of 11.5 months. A total of 180 patients (3.6%) were identified as experiencing rapid progression during a median follow-up of 9.9 months. The final multivariable analysis showed that younger age (P < .01), white race (P = .02), lower body mass index (P = .01), a diagnosis of peripheral arterial disease (P = .03), and a diagnosis of transient ischemic attack (P < .01) were associated with rapid progression.

Conclusions: Using a novel algorithm to extract data from >4 million patient records, we found that rapid progression of carotid stenosis appears to be rare. Although 17.6% of patients showed any degree of progression, only 3.6% had experienced rapid progression. Among those with any disease progression, 20.5% had experienced rapid progression. Although the overall incidence of rapid progression was low, patients with any progression might warrant close follow-up, especially if they have the associated risk factors for rapid progression. The custom software algorithm might be a powerful tool for creating and evaluating large datasets.

Keywords: Algorithm; Carotid; Natural language processing; Progression; Rapid; Stenosis; Stroke.