Pre-screening for non-diagnostic coronary computed tomography angiography

Area of Science:

• Cardiovascular Imaging
• Radiology
• Medical Informatics

Background:

• Indiscriminate referrals for coronary computed tomography angiography (CCTA) can lead to non-diagnostic studies.
• This results in increased resource utilization and delayed diagnosis for suspected coronary artery disease.

Purpose of the Study:

• To develop a simple clinical tool for predicting the likelihood of a non-diagnostic CCTA.
• To identify patients who might be better served by alternative diagnostic tests.

Main Methods:

• A clinical scoring system was developed using a cohort of 21,492 patients undergoing CCTA.
• Multivariable logistic regression and machine learning models were used to predict non-diagnostic CCTA likelihood.
• Predictor performance was validated using area under the curve (AUC).

Main Results:

• Both logistic regression and ML models showed fair discrimination (AUC 0.630 and 0.634, respectively).
• Key predictors of non-diagnostic studies included cardiac implants and weight >100 kg.
• The model achieved fair predictive ability for non-diagnostic CCTA outcomes.

Conclusions:

• A predictive model was developed for implementation at the point-of-scheduling.
• This tool can help identify patients best suited for alternative non-invasive diagnostic tests.
• The model aims to optimize diagnostic pathways and resource allocation.