Real world utilization of coronary artery calcium scoring in a large academic health system.

Peter A. Glynn: Writing – review & editing, Writing – original draft, Formal analysis, Data curation, Conceptualization. Reniell Iniguez: Writing – original draft, Validation, Data curation. Samuel Luebbe: Validation, Data curation. Philip Greenland: Writing – review & editing, Methodology, Conceptualization.

We conducted a retrospective observational study at Northwestern Medicine (NM), a large urban and suburban academic health system. We utilized the NM Enterprise Data Warehouse (EDW), a complete repository of clinical data from across the system, to identify patients aged 18 – 89 years-old who underwent CACS between January 2019 and May 2025. We excluded those who had CACS in conjunction with CT coronary angiography or PET/CT myocardial perfusion stress imaging. We extracted clinical features necessary to estimate ASCVD risk via the pooled cohort equations (PCE)[ 8 ] and collected additional information about risk enhancing factors- inflammatory arthritis, human immunodeficiency virus (HIV), hepatitis C, chronic kidney disease, end-stage kidney disease, family history of coronary artery disease (CAD), metabolic syndrome, South Asian ancestry, elevated lipoprotein(a) > 125 nmol/L and triglycerides > 200 mg/dL, and female specific risk factors such as pre-eclampsia, gestational hypertension and diabetes, polycystic ovarian syndrome, and endometriosis. Laboratory values were directly extracted, and relevant ICD 10 codes were used for clinical diagnoses. Because there is no specific ICD code for family history of premature CAD in a first degree relative, a more inclusive definition of any family history of CAD was used. Medications including current statin, aspirin, and anti-hypertensive use were extracted. All data were limited to those available up to the time the CACS was ordered. We manually extracted data from a random sample of 100 individuals with 97.8 % agreement in data needed to estimate ASCVD risk and 97.6 % agreement and ASCVD risk categorization between the EDW data and our manual review. We assessed deviations from current clinical practice guidelines as follows: presence of an active statin prescription prior to CACS, presence of prior DM diagnosis, presence of a prior CACS in the preceding 3 years, low (20 %) ASCVD risk estimated by the PCE, and young ( 80 years-old) age. We assessed the proportion of studies with at least one guideline deviation, stratifying results by self- or physician-referral status. We compared the proportion of inappropriate orders among these stratifications using Chi-square testing. As guidelines consider it may be appropriate to assess CACS in young individuals and those at low risk when risk enhancing factors are present, we assessed the proportion of young and low-risk individuals who had risk enhancing factors present. These scores were subsequently added to the appropriate total. All statistical analyses were performed using R version 3.1.2.

The NM EDW returned 43,271 consecutive CACS between January 2019 and May 2025. Of these, 9754 were excluded for incomplete data, leaving 33 , 517 CACS for analysis. The most common reasons for exclusion were missing total cholesterol and/or high-density lipoprotein cholesterol levels (6194) or missing data on race/ethnicity (3747), which frequently co-occurred. Sample characteristics are displayed in Fig. 1 A. Over 42 % of CACS were obtained in individuals at low ASCVD risk (20 %). Over 26 % of patients already had statins prescribed and 10 % had diabetes. Overall, approximately 28 % of CACS were obtained in individuals without any guideline deviations. When including young and low risk individuals with at least one risk enhancing factor and no other guideline deviations, the number of scans without a guideline deviation rose to 49 %. Approximately 36 % of individuals <40 years-old and 50 % of low-risk individuals had no risk enhancing factor present ( Fig. 1 B). While statistically significant differences existed between those who self-referred and those who were physician-referred, ASCVD risk distribution ( Fig. 1 C) and other findings were generally similar. The overall percentage of scans without a guideline deviation was significantly lower (44 %) in the self-referral group compared to the physician-referred group (52 %). Fig. 1 Distribution of clinical characteristics, guideline adherence, risk enhancing factors, and atherosclerotic cardiovascular risk among adults undergoing coronary artery calcium assessment. Fig 1: dummy alt text Distribution of clinical characteristics, guideline adherence, risk enhancing factors, and atherosclerotic cardiovascular risk among adults undergoing coronary artery calcium assessment.

In this large, contemporary health-system analysis of >33,000 CAC scans, a substantial proportion of tests were obtained in individuals whose clinical profiles fell outside traditional guideline-based indications. Our results align with prior studies which reported that a large proportion of CAC testing occurred among low- and high-risk individuals, including those already prescribed preventive therapy [ 9 ]. CACS appears to have a motivational patient benefit as higher CAC scores are associated with greater initiation of preventive therapies and lifestyle changes [ 10 ]. Consequently, guideline deviations we observed may reflect a shift in practice whereby CAC tests are used not as the primary determinant of statin eligibility, but as a tool to strengthen patient engagement and reinforce adherence to cardioprotective behaviors or statins. Understanding the broader contextual factors that influence CAC ordering, such as how clinicians and patients use CACS to navigate uncertainty, negotiate preventive therapy adherence, or engage in shared decision-making, may clarify the drivers behind observed utilization patterns.

Several limitations should be considered. First, analyses were restricted to structured EHR data available at the time of test ordering; we did not assess clinician reasoning, patient preferences, or decision-making processes that may justify CAC use beyond guideline indications. There are some inherent limitations to structured EHR data, which require active input providers (and thus some conditions may be missed or documented incorrectly). Some ICD 10 codes lack nuance, for instance there is not a code specifically for premature history of CAD in a first-degree relative. However, on manual review of a random sample of charts, we found excellent agreement between the EHR data and manual review and risk estimation. Next, clinical outcomes following CAC testing were not tracked. As such, no comments can be made whether deviation from guidelines resulted in a positive or negative effect. A significant subset of patients lacked complete lipid measurements and race/ethnicity data, resulting in their exclusion; although necessary for ASCVD risk calculation, this may have introduced selection bias. Our findings reflect practices within a single academic health system and may not generalize to other settings with different referral pathways, patient populations, or preventive cardiology practices. Finally, our measure of guideline deviation is based on established risk categories and recognized indications but cannot fully capture the nuanced clinical scenarios in which CAC may reasonably be used.

Coronary artery calcium scoring (CACS) by computed tomography (CT) is a widely used clinical tool, where CACS demonstrate strong, graded positive associations with future cardiovascular events [ [1] , [2] , [3] , [4] ]. American College of Cardiology(ACC)/American Heart Association(AHA) guidelines recommend considering CACS in individuals aged 40–75 years-old without diabetes mellitus and an LDL- C ≥ 70 mg/dl-189 mg/dL with borderline (10-year atherosclerotic cardiovascular disease (ASCVD) risk 5.0 to <7.5 %) or intermediate risk (≥7.5–20 %) when the decision regarding statin therapy initiation is uncertain, or in selected young (<40 years-old) or low risk individuals (10-year ASCVD risk < 5 %) when risk enhancing factors are present [ 5 ]. When a CACS is zero, repeat assessment after approximately 5 years may be considered in selected individuals, consistent with National Lipid Association guidance; earlier rescanning is generally not recommended [ 6 ]. In addition, the 2022 ACC Expert Consensus Decision Pathway on the Role of Non-Statin Therapies incorporates CAC burden to define LDL-C thresholds for treatment intensification, including the addition of non-statin agents in selected primary prevention populations [ 7 ]. Despite widespread availability, little is known about real-world utilization of CACS. Many health systems now allow patients to self-refer for CACS, and little is known about the impact of this practice on guideline adherence. Accordingly, we examined real-world CACS utilization in a large academic health system to assess current practice patterns and their relationship to guideline recommendations.

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Philip Greenland reports a relationship with 10.13039/100000002 National Institutes of Health that includes: funding grants. Philip Greenland- Senior Editor, Journal of the American Medical Associaion If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.