Objective:

This study aimed to assess the proportion of CTA performed among U.S. Medicare beneficiaries in both stroke and non-stroke cohorts and characterize its variation at the hospital-level.

Background:

Non-invasive imaging of carotid and intracranial vessels (referred to as "CTA") is used for diagnosis, targeting secondary prevention, and screening for large vessel occlusion in stroke patients. While CTA has clear utility, it also has potential downsides including radiation exposure, costs and incidental findings/false positives. 

Design/Methods:

We conducted a cross-sectional analysis using 2021 Medicare data. CTA was identified using CPT codes in a 5% Carrier file. Claims were linked to Inpatient and ED Outpatient data to obtain detailed clinical context for each visit. Patients were determined as stroke or non-stroke cohorts based on ICD-10-CM diagnosis codes during their CTA visit. We used multi-level linear regression models, clustered at hospital level and adjusted for transfer-in/transfer-out status, to estimate the "shrunken mean" of CTA rates among stroke and non-stroke patients for each hospital.

Results:

We identified 3090 hospitals in the study. Shrunken mean CTA rates for stroke patients varied substantially, from 24.6% to 86.3% with a median of 56.7% [IQR 50%-65.3%]. CTA use also varied for non-stroke patients, ranging from 1.7% to 18.1% with median of 4.4% [IQR 3.7-5.7]. Stroke and non-stroke imaging rates were correlated (correlation coefficient=0.4). However, there was significant variation, of the 1560 hospitals with above median CTA rates for stroke diagnoses, 37.3% had below median rates of CTA use for non-stroke diagnoses

Conclusions:

CTA was widely used for both stroke and non-stroke patients but varied significantly at the hospital level. This variation may represent an opportunity to optimize CTA use. The existence of hospitals with relatively high CTA use in stroke patients and low use in non-stroke patients suggests it is possible to apply CTA selectively.