Preventive management of non-obstructive coronary artery disease after coronary CT angiography in the Emergency Department

Abstract

Objectives:

We sought to assess medical management of patients found to have non-obstructive coronary artery disease (CAD) on coronary computed tomography angiography (CCTA) in the Emergency Department (ED).

Background:

Contemporary recognition and management of non-obstructive CAD discovered on ED-performed CCTA is unknown.

Methods:

We considered patients undergoing CCTA in our hospital’s ED between November 2013 and March 2018 who also received primary care within our health system. We included all patients with non-obstructive CAD, defined as 1–49% maximum luminal stenosis on CCTA, and a control group without CAD, 1 case:1 control. Ten-year atherosclerotic cardiovascular disease (ASCVD) risk prior to CCTA was estimated using the Pooled Cohort Equations. Management changes were recorded until 6 months after CCTA. Multivariable logistic regression tested the association of CCTA result with follow-up statin prescription, adjusting for cardiovascular risk factors and baseline statin use.

Results:

Our cohort included 510 patients with non-obstructive CAD and 510 controls. Prevalence of statin prescription increased from 38.8% to 56.1% among patients with non-obstructive CAD (p < 0.001) and 18.0% to 20.4% among controls (p = 0.01), representing a 7.1-fold relative difference (95% CI 4.4–23.0; p < 0.001) in multivariable analysis. However, 30.0% of patients with non-obstructive CAD and ≥20% 10-year ASCVD risk were not prescribed a statin at the end of follow-up. Cardiologist evaluation was independently associated with statin prescription after adjustment for ASCVD risk factors (OR 4.4; 95% CI 2.4–8.5; p < 0.001). A CAD-RADS 1–2 result was associated with lower low-density lipoprotein cholesterol by 12.1 mg/dL at mean 1.9-year follow-up (p < 0.001).

Conclusions:

Incidental subclinical atherosclerosis on ED-performed CCTA increases the likelihood of statin prescription, but opportunities to improve allocation of indicated preventive therapies remain.

Introduction

Coronary computed tomographic angiography (CCTA) is increasingly used for the evaluation of chest pain in the Emergency Department (ED) to non-invasively diagnose coronary artery disease (CAD).^1–7 Approximately 10–15% of individuals who undergo CCTA for suspected CAD in the ED are found to have obstructive CAD in randomized controlled trials.^2,4,8 However, approximately 50% are discovered to have non-obstructive CAD, which is independently associated with increased future cardiovascular disease risk.⁹ The 2018 ACC/AHA cholesterol practice guidelines support the use of cardiac CT-derived measures of subclinical coronary atherosclerosis to refine statin suitability when estimated 10-year atherosclerotic cardiovascular disease (ASCVD) risk is greater than 5% in the primary prevention setting and there is statin prescription equipoise.¹⁰ As most health insurers to date do not cover the use of cardiac CT to assess subclinical coronary atherosclerosis in asymptomatic individuals, opportunistic detection in the evaluation of acute chest pain has emerged as a pathway to improve primary prevention of CAD.

Within clinical trials of CCTA to evaluate stable chest pain in the outpatient setting, as well as observational studies of outpatient CCTA, CCTA use is strongly associated with new prescriptions of statins.^5,11–16 Compared to the standard-of-care, CCTA use within the SCOT-HEART trial was associated with higher rates of statin initiation and a reduction of the composite endpoint of CAD death and non-fatal myocardial infarction at 5 years.¹⁷ Preventive therapy escalation by the ordering longitudinal clinician for patients at heightened ASCVD risk may be one mechanism of the apparent clinical benefit of CCTA.^9,18 However, statin escalation prompted by ED-performed CCTA in clinical trials is less readily apparent.^8,11,19–21 Whether opportunistic subclinical atherosclerosis detection from ED-performed CCTA in current clinical practice prompts either inpatient or longitudinal clinicians to initiate statins remains largely unknown.

Here, we sought to assess the contemporary management of non-obstructive CAD discovered by CCTA in the ED outside of a clinical trial setting. We hypothesized that the presence of non-obstructive CAD, compared to the absence of CAD, would be associated with an increased likelihood of statin prescription at follow-up. Furthermore, we explore opportunities to extend primary prevention from the ED to the outpatient setting.

Methods

Study population

In this case-control study, all patients undergoing CCTA to evaluate a complaint of acute chest pain or possible acute coronary syndrome in the ED at Massachusetts General Hospital between November 1, 2013, and March 31, 2018, were considered for inclusion. Massachusetts General Hospital is an academic tertiary care hospital in Boston, Massachusetts, and its ED treats more than 100,000 patients annually. To ensure accurate assessment of management changes following CCTA, we restricted our analysis to patients who were receiving primary care, or who newly established primary care within 6 months of CCTA, in the hospital’s parent healthcare system, Partners Healthcare. Partners Healthcare is a network of academic and community hospitals and outpatient centers serving eastern Massachusetts with a shared electronic medical record. Primary care provider assignments were verified by manual chart review.

All patients meeting the above criteria with non-obstructive coronary artery disease, defined as Coronary Artery Disease Reporting and Data System (CAD-RADS) classification²² 1–2 (i.e., 1–49% maximum luminal stenosis) on CCTA, were included as “cases.” Patients with obstructive CAD (≥50% maximum luminal stenosis, i.e., CAD-RADS classification 3–5) were excluded. To study the association of a CAD-RADS 1–2 result with subsequent preventive management, we created a control group found to have no CAD (i.e., CAD-RADS classification 0) on ED-performed CCTA. Controls were matched by age within 1 year and sex, 1 case to 1 control, until matched controls were exhausted; at this point, patients of the same sex were matched irrespective of age until sex matches were exhausted; finally, patients of the opposite sex with the closest available age were selected until case and control groups were equal in size. If a patient underwent CCTA more than once during the study period, only the first CCTA was included. The Partners Healthcare Institutional Review Board approved this study.

Coronary CT angiography

All patients were scanned on 128-slice or 192-slice dual-source CT scanners (SOMATOM Definition Flash or Somatom Force respectively, Siemens Healthineers, Forcheim, Germany), as described previously.²³ Non-contrast coronary artery calcium score CT was performed in all patients with a prospectively ECG-triggered protocol. CCTA was performed with a prospective or retrospective ECG gating depending on patient factors as determined by the supervising physician. Beginning in September 2016, standard CAD-RADS terminology was included in radiology reports per Society for Cardiac Computed Tomography guidelines.²² Prior to September 2016, a similar CCTA reporting system was used at Massachusetts General Hospital, and reports were converted to equivalent CAD-RADS categories for the purposes of this analysis. In our cohort, 44% of CT studies were performed after the adoption of standardized CAD-RADS reporting. Reporting templates before and after the adoption of CAD-RADS are provided in the Supplementary Appendix. Clinical radiology reads were used to assign CAD-RADS categories.

Data extraction

Patients’ demographic information, cardiovascular risk factors, laboratory testing, and medication use prior to undergoing CCTA were extracted from the medical record by manual chart review. We defined diabetes mellitus as hemoglobin A1c ≥6.5%, or documented diagnosis or treatment of diabetes using relevant medications. We defined hypertension as a chart diagnosis of hypertension or use of antihypertensive medications. The last ambulatory blood pressure prior to ED presentation was recorded; if no prior blood pressure was available, the blood pressure at ED presentation was used. Additionally, the most recent lipid panel prior to ED presentation was recorded. These data (all of which were available to clinicians prior to knowledge of the CCTA result) were used to calculate a baseline 10-year risk of ASCVD the Pooled Cohort Equations (PCE).²⁴ Variables to calculate baseline PCE risk were available in 453 of 510 (89%) CAD-RADS 0 patients and 483 of 510 (95%) CAD-RADS 1–2 patients. Subsequent inpatient and outpatient management was recorded until 6 months after CCTA. Additionally, to examine associations between CCTA result and longer-term lipid trajectories, follow-up lipid levels were extracted up to March 31, 2018, or 6 months after ED CCTA, whichever date occurred later. The primary end-point was a statin medication prescription at the end of follow-up.

Statistical analysis

Continuous variables were compared using the Student’s t-test, and categorical variables were compared using the Pearson’s chi-squared test or Fisher’s exact test as appropriate. Changes in medication prescribing before and after CCTA were compared using McNemar’s test. The Cochran-Armitage test was used to analyze statin prescribing trends across risk categories.

Multivariate logistic regression was used to assess the association between CCTA results and statin prescription, adjusting for baseline statin use and ASCVD risk factors. Additionally, multivariable linear regression models were used to test changes in low-density lipoprotein (LDL) levels before and after CCTA. P-values less than 0.05 were considered statistically significant. Analyses were performed using R 3.3 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Baseline patient characteristics

Our final cohort included 510 patients with non-obstructive CAD (i.e., CAD-RADS 1–2 on CCTA) and 510 patients without CAD (i.e., CAD-RADS 0). Patients with non-obstructive CAD were older (mean age 58.8 vs. 52.8 years, p < 0.001), were more likely to be male (53.3% vs. 44.5%, p = 0.006), and had higher BMI (30.3 kg/m² vs. 29.1 kg/m², p = 0.007, Table 1). Patients with non-obstructive CAD had a higher prevalence of hypertension (53.3% vs. 41.5%, p < 0.001), diabetes mellitus (17.8% vs. 9.6%, p = 0.007), past or current smoking (31.0% vs. 23.9%, p = 0.01), and hemoglobin A1c ≥7.0% at baseline (8.4% vs. 3.3%, p = 0.009). Baseline LDL cholesterol levels prior to CCTA were available in 92% of the cohort and did not differ significantly between groups (mean (SD) LDL cholesterol 115.7 (34.8) mg/dL in patients with non-obstructive CAD vs. 121.0 (94.1) mg/dL in patients without CAD, p = 0.26). The baseline statin prescription rate was 18.0% in patients without CAD and 38.8% patients with non-obstructive CAD (p < 0.001).

Table 1.

Patient baseline characteristics by CAD-RADS category (N [%] or mean ± SD).

	CAD-RADS 0	CAD-RADS 1–2	P-value
N	510	510
Age, in years	52.8 ± 10.2	58.8 ± 10.0	< 0.001
Female	283 (55.5%)	238 (46.7%)	0.006
Race/ethnicity			0.02
• Asian	19 (3.7%)	15 (2.9%)
• Black	64 (12.5%)	35 (6.8%)
• Hispanic	65 (12.7%)	58 (11.3%)
• White	333 (65.3%)	371 (72.7%)
• Other	29 (5.7%)	31 (6.1%)
BMI, in kg/m2	29.1 ± 6.1	30.3 ± 7.0	0.007
Systolic blood pressure, in mmHg*	127 ± 16	128 ± 18	0.17
Diastolic blood pressure, in mmHg*	77 ± 10	78 ± 10	0.63
Hypertension	210 (41.2%)	272 (53.3%)	< 0.001
Hyperlipidemia	133 (26.1%)	158 (31.0%)	0.10
Diabetes mellitus	49 (9.6%)	91 (17.8%)	0.007
Ever smoking	122 (23.9%)	158 (31.0%)	0.01
Current smoking	25 (4.9%)	40 (7.8%)	0.07
Aspirin use at baseline	87 (17.1%)	193 (37.8%)	< 0.001
Aspirin allergy or intolerance	10 (1.4%)	16 (3.1%)	0.30
Statin use at baseline	92 (18.0%)	198 (38.8%)	< 0.001
Statin allergy or intolerance	11 (2.2%)	23 (4.5%)	0.055
Low-density lipoprotein cholesterol, in mg/dL*	121.0 ± 94.1	115.7 ± 34.8	0.26
Hemoglobin A1c, in %*	5.6 ± 0.9	5.9 ± 1.1	< 0.001
Hemoglobin A1c ≥7.0%*	17 (3.3%)	43 (8.4%)	0.009
Calculated 10-year atherosclerotic cardiovascular risk, in %	4.4 ± 3.9	10.6 ± 10.2	< 0.001

^*Last values measured/recorded in the outpatient setting prior to emergency department presentation

CAD-RADS = Coronary Artery Disease Reporting and Data System; BMI = body mass index

Mean (SD) calculated 10-year ASCVD risk using clinical risk factors measured prior to CCTA was 10.6 (10.2)% among patients with non-obstructive CAD and 4.4 (3.9)% among patients without CAD (p < 0.001). Thirty-five percent of patients with non-obstructive CAD had low (<5%) 10-year ASCVD risk, compared to 69% of patients without CAD (Figure 1). Thirty-six percent of patients with non-obstructive CAD had intermediate (≥7.5% and <20%) risk, and 14% had high (≥20%) risk. Two patients with a CAD-RADS 0 result (0.4%) had high (≥20%) 10-year ASCVD risk in our dataset.

Figure 1.

Distribution of atherosclerotic cardiovascular risk classification by CAD-RADS category.

Statin therapy after Emergency Department coronary computed tomography angiography

Among individuals with at least intermediate (≥7.5%) ASCVD risk, 41% had statin prescriptions at baseline; this was not different for those with CAD-RADS 0 versus 1–2 results (p = 1). Overall, the prevalence of statin prescription increased from 38.8% to 56.1% among patients found to have non-obstructive CAD on CCTA (p < 0.001), corresponding to a 44% relative increase, and from 18.0% to 20.4% among those without CAD (p = 0.01, Table 2). In a sparse model adjusted only for baseline statin use, a CAD-RADS 1–2 versus 0 result was associated with a 9.9-fold increase in statin prescription odds at follow-up (95% CI 5.9–17.6, p < 0.001). In a multivariable analysis, a 7.1-fold increase in statin prescription odds (95% CI 4.4–23.0; p < 0.001) was observed after a CAD-RADS 1–2 result, adjusting for baseline statin use, age, sex, diabetes, hypertension, current smoking, race, and evaluation by a cardiologist in the ED or inpatient setting and/or in the outpatient setting (Table 3, Central Illustration). The association between CAD-RADS result and statin prescription did not differ between patients who underwent CCTA before or after implementation of standardized CAD-RADS reporting in September 2016 (p = 0.85).

Table 2.

Patient management before and after coronary CT angiography by CAD-RADS category.

	CAD-RADS 0 (n = 510)	CAD-RADS 1–2 (n = 510)	P-value
Lipid surveillance
Lipids checked within 2 previous years	348 (68.2%)	402 (78.8%)	< 0.001
Lipids checked within 6 months after CCTA	177 (34.7%)	266 (52.2%)	< 0.001
Statin prescription and discontinuation
Statin use at baseline	92 (18.0%)	198 (38.8%)	< 0.001
Statin use at the end of study period	104 (20.4%)	286 (56.1%)	< 0.001
Statin initiated in the hospital	4 (0.8%)	23 (4.5%)	< 0.001
Statin initiated in the outpatient setting	11 (2.2%)	67 (13.1%)	< 0.001
Statin dose increased in the hospital or outpatient setting	8 (1.6%)	36 (7.0%)	m< 0.001
Providers starting or increasing statin			0.36
• ED/inpatient	8 (1.6%)	31 (6.1%)
• Outpatient primary care	13 (2.5%)	69 (13.5%)
• Outpatient specialist	2 (0.4%)	26 (5.1%)
Statin dose decreased	0 (0%)	1 (0.2%)	1
Statin discontinued	3 (0.6%)	2 (0.4%)	1
Providers decreasing or discontinuing statin			1
• ED/inpatient	1 (0.2%)	0 (0%)
• Outpatient primary care	1 (0.2%)	2 (0.4%)
• Outpatient specialist	1 (0.2%)	1 (0.2%)
Involvement by cardiologists
Inpatient cardiology consultation	32 (6.3%)	60 (11.8%)	0.003
Outpatient cardiology evaluation	82 (16.1%)	144 (28.2%)	< 0.001
Inpatient and/or outpatient cardiology evaluation	92 (18.0%)	165 (32.4%)	< 0.001
Other management
Non-statin LDL-lowering medications before and after CCTA	4 (0.8%)	7 (1.4%)	0.55
Aspirin use at baseline	87 (17.1%)	193 (37.8%)	< 0.001
Aspirin use at the end of the study period	105 (20.3%)	264 (51.8%)	< 0.001
Nutrition referral	5 (1.0%)	20 (3.9%)	0.004
Smoking cessation pharmacotherapy and/or consultation	13 (2.5%)	13 (2.5%)	1
Primary care follow-up within 6 months	417 (81.8%)	457 (89.6%)	< 0.001
Time to primary care follow-up, days	80.1 ± 122.8	56.0 ± 97.9	< 0.001
Primary care and/or cardiology follow-up within 6 months	421 (82.5%)	461 (90.4%)	< 0.001

CAD-RADS = Coronary Artery Disease Reporting and Data System; CCTA = cardiac computed tomography angiography; ED = emergency department

Table 3.

Model outputs for predictors of statin prescription at follow-up.

Variable	Odds ratio	95% confidence interval	P-value
CAD-RADS 1–2 (compared to CAD-RADS 0)	7.1	4.4–23.0	< 0.001
Age (per year)	1.04	1.0–1.1	0.01
Male sex (compared to female sex)	0.9	0.5–1.7	0.70
Black race (compared to non-black race)	0.4	0.1–1.3	0.17
Diabetes	1.2	0.5–2.8	0.65
Current smoking	2.4	0.8–6.3	0.10
Hypertension	0.7	0.4–1.3	0.24
Evaluation by a cardiologist in the inpatient or outpatient setting	4.4	2.4–8.5	< 0.001
Statin use at baseline	705.7	232.4–2,829.0	< 0.001

CAD-RADS = Coronary Artery Disease Reporting and Data System

Central Illustration.

The finding of non-obstructive coronary artery disease on coronary CT angiography in the Emergency Department is associated with higher likelihood of statin prescription and lower low-density lipoprotein levels at follow-up. ED = Emergency Department; CCTA = coronary CT angiography; CAD-RADS = Coronary Artery Disease Reporting and Data System; LDL = low-density lipoprotein.

Statin therapy was newly initiated in 90 patients with non-obstructive CAD and 15 patients without CAD (17.6% vs. 2.9%, p < 0.001). Additionally, among patients already taking statins, the statin dose was increased for 36 (7.0%) of patients with non-obstructive CAD and for 8 (1.6%) of patients without CAD (p < 0.001). Across the cohort, statins were initiated or escalated in 149 patients within 6 months of CCTA, and outpatient clinicians provided 110 (74%) of these prescriptions. Three patients without CAD (0.6%) and 3 patients with non-obstructive CAD (0.6%) had statins decreased or discontinued after ED CCTA (p = 1).

Among those with a CAD-RADS 0 result, there was a small increase in statin prescriptions after CCTA (Figure 2a). At the end of follow-up, 14.5% of CAD-RADS 0 patients with low (<5%) 10-year ASCVD risk were prescribed statin therapy. Among those with a CAD-RADS 1–2 result, post-test statin prescription increases were consistent across 10-year estimated ASCVD risk categories (p for trend = 0.7) (Figure 2b). At the end of follow-up, however, 30% of CAD-RADS 1–2 patients at high (≥20%) calculated 10-year ASCVD risk were not prescribed statins, and 33% of CAD-RADS 1–2 patients at intermediate (≥7.5% and <20%) 10-year ASCVD risk were not prescribed statins. Notably, 90.4% of patients with non-obstructive CAD saw a primary care provider and/or cardiologist in outpatient follow-up within 6 months of CCTA.

Figure 2.

Prevalence of statin prescription before coronary CT angiography and at the end of follow-up for (a) patients with a CAD-RADS 0 result, (b) patients with a CAD-RADS 1–2 result.

Predictors of statin prescription

Among patients found to have non-obstructive CAD, 68% of patients with coronary artery calcium (CAC) score >100 were prescribed a statin at the end of follow-up, compared with 34% of patients with CAC ≤100 (p < 0.001). Across bins of CAC score, the proportional increase in statin prescription at follow-up was greater with increasing CAC score (p trend < 0.001, Supplementary Figure S1). Prevalence of statin prescription increased non-significantly among patients with baseline LDL cholesterol ≥190 mg/dL (n = 22) from 41% to 64% after CCTA (p = 0.07), increased among those with diabetes mellitus (n = 140) from 64% to 69% (p = 0.04), and increased among active smokers found to have non-obstructive CAD (n = 40) from 30% to 58% (p = 0.003).

In a multivariable model evaluating the association of a CAD-RADS 1–2 versus 0 result with statin prescription, older age, baseline statin prescription, and cardiologist evaluation were at least nominally independently associated (p < 0.05) with statin prescription during follow-up (Table 3). Cardiologist evaluation in the hospital or in the outpatient setting was independently associated with statin prescription after adjustment for ASCVD risk factors (OR 4.4; 95% CI 2.4–8.5; p < 0.001). Among the subset of patients with a CAD-RADS 1–2 result, age, cardiologist evaluation, and baseline statin use were significantly associated with statin prescription at follow-up (Supplementary Table S1). Among patients with a CAD-RADS 1–2 result and ≥7.5% 10-year ASCVD risk, cardiologist evaluation and baseline statin use were associated with statin prescription (Supplementary Table S2). Comparing patients who received new or intensified statin prescriptions vs. others (Supplementary Table S3), patients with new or intensified statin prescriptions were older (mean (SD) age 58.9 (8.9) vs. 55.3 (10.7) years, p < 0.001), were more likely to be current smokers (11.4% vs. 5.5%, p = 0.01), were less likely to have black race (4.7% vs. 10.6%, p = 0.04), had higher 10-year ASCVD risk (mean (SD) 10.5 (9.0)% vs. 7.1 (8.1)%, p < 0.001), were more likely to have been seen by a cardiologist (49.7% vs. 21.0%, p < 0.001), and were more likely to have had primary care follow-up within 6 months (98.6% vs. 87.9%, p < 0.001).

Among individuals with non-obstructive CAD, patients prescribed statins were older and had higher rates of diabetes mellitus, hypertension, non-black race, cardiologist evaluation, and primary care follow-up within 6 months compared to patients not prescribed statin therapy (Supplementary Table S3). Among patients with non-obstructive coronary artery disease and high (≥20%) 10-year ASCVD risk, patients who did not receive statins were more likely than those receiving statins to have black race (19.0% vs. 2.0%; p = 0.03; Supplementary Table S4). All 5 of the black patients in our cohort with high 10-year ASCVD risk had primary care follow-up within 6 months. None of these 5 patients were evaluated by a cardiologist during the study period.

Change in low-density lipoprotein

The baseline LDL cholesterol level prior to CCTA was available for 928 (91%) patients, and a follow-up LDL cholesterol after CCTA was available for 773 (76%) patients. Mean (SD) baseline LDL cholesterol was 121.0 (94.1) mg/dL vs. 115.7 (34.8) mg/dL in patients with a CAD-RADS 0 vs. 1–2 result, respectively (p = 0.26). The last available mean (SD) follow-up LDL cholesterol level, measured at a mean (SD) 1.9 (1.1) years after CCTA, was 116.8 (35.2) mg/dL in patients without CAD and 101.9 (36.6) mg/dL in patients with non-obstructive CAD (p < 0.001). At follow-up, patients with a CAD-RADS 1–2 result were more likely than those with a CAD-RADS 0 result to have an LDL cholesterol level <70 mg/dL (19.7% vs. 8.4%, p < 0.001) and more likely to have an LDL cholesterol level <100 mg/dL (52.1% vs. 32.0%, p < 0.001, Figure 3).

Figure 3.

Distribution of low-density lipoprotein cholesterol levels by CAD-RADS category at a mean 1.9 years after undergoing coronary CT angiography.

Among patients newly initiated on statin therapy within 6 months after CCTA, mean (SD) LDL cholesterol was 131.7 (38.9) mg/dL prior to CCTA and 97.1 (40.0) mg/dL at last available follow-up (p < 0.001). In multivariable linear regression adjusted for age, sex, and baseline LDL cholesterol level, a CAD-RADS 1–2 result was associated with a lower LDL cholesterol level at follow-up by 12.1 mg/dL (p < 0.001) compared to a CAD-RADS 0 result. Among patients not prescribed a statin at 6 months after CCTA, mean (SD) LDL cholesterol was 119.0 (30.9) mg/dL at baseline and 119.4 (32.8) mg/dL at last available follow-up (p = 0.33).

Additional preventive management measures before and after CCTA

Seven (1.4%) patients with non-obstructive CAD and 4 (0.8%) of patients without CAD were taking a non-statin LDL cholesterol-lowering medication (ezetimibe and/or a fibrate) prior to CCTA (p = 0.55); none of these medications were changed in follow-up, and no new non-statin LDL cholesterol-lowering prescriptions were provided in follow-up. No patients were prescribed a PCSK9 inhibitor during the study period.

Aspirin prescriptions increased from 37.8% to 51.8% among patients with non-obstructive CAD (p < 0.001) and from 17.1% to 20.3% (p < 0.001) among those without CAD (p = 0.48) (Table 2; Supplementary Figure S2). In a multivariable model adjusted for baseline aspirin use, age, sex, race, diabetes, hypertension, current smoking, and cardiologist evaluation, a CAD-RADS 1–2 result was associated with 4.2-fold increased likelihood of aspirin prescription at follow-up (95% CI 2.2–8.5, p < 0.001) compared to a CAD-RADS 0 result. Evaluation by a cardiologist in the ED/inpatient and/or outpatient setting was associated with 3.6-fold greater odds of aspirin prescription (95% CI 1.9–6.7, p < 0.001).

Of patients with a CAD-RADS 1–2 result and BMI ≥35 (n = 105), 9 (8.6%) were referred to a nutritionist within 6 months. Additionally, of the patients with non-obstructive CAD who were actively smoking tobacco, 13 (33.3%) were prescribed pharmacotherapy for smoking cessation or referred for a dedicated smoking cessation consultation.

Discussion

In this study, the discovery of non-obstructive CAD on CCTA performed in the ED was strongly associated with higher likelihood of statin prescription at 6 months. Opportunistic detection of non-obstructive CAD by ED CCTA was associated with significant improvement in the provision of statins to statin-eligible patients for primary prevention, typically by longitudinal outpatient clinicians, within 6 months following the CCTA with resultant lower LDL cholesterol values at mean 2-year follow-up. Nevertheless, 1 in 3 individuals at intermediate-to-high short-term ASCVD based on clinical risk factors as well as radiographic coronary atherosclerosis remained without statins, despite >90% of patients following up with a primary care provider and/or cardiologist within 6 months of the abnormal CCTA.

To our knowledge, this is the first study to examine both ED and long-term outpatient management changes after incidental non-obstructive CAD is detected by ED-performed CCTA in real-world clinical practice. An analysis within the Rule Out Myocardial Infarction using Computed Angiographic Tomography (ROMICAT)-II trial examined medication changes after ED-performed CCTA through discharge from the index ED visit/hospitalization and found that 31% of subjects with non-obstructive CAD were prescribed a statin at discharge.²⁰ In an analysis of the ACRIN trial comparing ED-performed CCTA to standard of care in the evaluation of acute chest pain, Chang et al. reported risk ratios for statin prescription of 1.64 and 1.86 at 30 days and 1 year after ED presentation, respectively, for patients with 1–49% luminal stenosis on CCTA (n = 120) compared to patients randomized to standard of care not including CCTA (i.e., no anatomic assessment).²¹ The present work extends these prior studies by incorporating calculated a priori ASCVD risk and detailed outpatient follow-up in a larger, contemporary CCTA cohort external to a clinical trial. The observed changes in statin prescriptions after imaging strongly suggest that that longitudinal providers, who prescribed three-quarters of new statin prescriptions and dose escalations, recognized and altered management based on CCTA findings. Further, we found that cardiologists were more likely than non-cardiology clinicians to prescribe statin therapy. This finding is consistent with that of a prior registry-based study, which found that patients seen by cardiologists were more likely than others to receive guideline-directed statin therapy in both primary and secondary prevention contexts.²⁵ Whether this trend reflects greater recognition of long-term clinical risk associated with subclinical atherosclerosis or greater risk among patients referred to cardiologists is challenging to disentangle. Nevertheless, we identify greater prescriptions with cardiologist evaluation independent of conventional cardiovascular risk factors.

Our study may have important implications for ASCVD risk refinement and improving the allocation of preventive medicines through existing clinical workflows. First, the incidental finding of subclinical coronary atherosclerosis by ED-performed CCTA to evaluate potential acute coronary syndrome appears generally recognized as actionable. Within the SCOT-HEART trial, the presence of non-obstructive CAD triggered a recommendation to treating clinicians to intensify preventive therapies, which may have been responsible for the apparent benefit observed with CCTA.^2,17 Without such a protocol in place, we still observe escalation of preventive medicines. However, 1 in 3 patients at either intermediate or high 10-year ASCVD risk were still not initiated on statins despite risk confirmation with radiographic atherosclerosis. Practice was likely influenced by the 2013 ACC/AHA cholesterol practice guidelines, wherein CAC presence is a IIb level recommendation for supporting statin prescription among then-intermediate (5–7.5%) risk patients. While strength of recommendation for CAC use is now increased in the 2018 guidelines to IIa among now-intermediate risk patients (7.5–20%) to guide statin prescriptions as necessary, the 2013 guidelines more strongly supported statin prescriptions for individuals with ≥7.5% risk. Thus, while evolving guidelines are likely to change practice patterns, incidental coronary atherosclerosis may further inform statin prescription discussions in the primary prevention setting, even when incidentally detected in the evaluation of chest pain. Looking ahead, this gap may be bridged by leveraging reporting and communication systems to optimize use of preventive medicines where appropriate. A reporting system that harmonizes radiology, ED, and outpatient workflows would ideally integrate both validated clinical ASCVD risk scores and radiographic findings. Furthermore, since we observed that cardiology engagement was associated with improved allocation of statins, remote consultation (“e-consultation”) as previously employed within our practice may facilitate risk refinement and prevention recommendations.²⁶ We also observed low use of preventive interventions for obesity and smoking, which may also be optimized through electronic systems.

Second, as statin prescriptions increased similarly across the spectrum of ASCVD risk when non-obstructive CAD was detected, CCTA findings and clinical risk estimation are not well integrated. According to current guidelines, statins are reasonable when risk is intermediate and CAC score is greater than 100, and should be further individualized when CAC score is 1–99.¹⁰ However, CCTA findings in patients at low estimated 10-year ASCVD risk and the presence of non-calcified plaque (i.e., CAD-RADS 1–2 but CAC score 0) are not addressed in these guidelines. Whether an observation of CAD-RADS 1–2 should trump low 10-year ASCVD clinical risk scoring or CAC score of 0 requires a more individualized approach as well as further study.

Third, the absence of subclinical coronary atherosclerosis on ED CCTA does not prompt de-escalation of statins, even when estimated ASCVD risk is low based on clinical risk factors, although this is a new consideration of the 2018 ACC/AHA cholesterol guidelines. In our cohort, 20% of patients with a CAD-RADS 0 result were prescribed statin therapy, including 15% of those with <5% 10-year ASCVD risk. The absence of CAC, particularly when confirming low estimated 10-year ASCVD risk, is associated with a very low rate of adverse cardiovascular events.^27,28 Recent work with nearly 10 years of follow-up suggests reduced benefit of statin therapy for primary prevention in people with a CAC score of zero.²⁹ The 2018 ACC/AHA cholesterol guidelines support the use of CAC score 0 to withhold statins based on epidemiologic studies but do not address potential de-escalation.^28,30,31 Although the safety of withdrawing statins using CAC score 0 requires further study, CAC score 0 from ED-performed CCTA could be used to consider statin de-escalation in non-smoking patients with low ASCVD risk.

Fourth, we found that aspirin prescriptions increased in tandem with statin prescriptions in patients with a CAD-RADS 1–2 result. The use of aspirin is widely accepted in the context of secondary prevention,³² but its role in primary prevention is less clear and has been called into question by a series of recent trials (published after our study period) with negative or equivocal results,^33–35 resulting in downgraded recommendations for the use of primary prevention aspirin in the 2019 ACC/AHA primary prevention guidelines.³⁶ Whether efficacy is present among a subset with subclinical coronary atherosclerosis already on statins requires dedicated studies.

While our study has several strengths, key limitations should be considered. First, to ensure complete follow-up and access to detailed clinical information, we restricted analysis to patients receiving longitudinal care in our health system. It is possible that patients obtaining longitudinal care within a different health system have a lower rate of follow-up statin prescriptions when atherosclerosis is identified due to non-integration of health record systems. Second, this study involved an academic medical center with considerable CCTA experience. Nevertheless, our center’s high volume of CCTA allows for a larger contemporaneous sample of individuals with non-obstructive CAD detected in the ED. Finally, we were unable to capture patient preferences from shared decision-making discussions, which may have affected longitudinal management decisions. However, reported statin intolerance was rare and did not significantly differ between groups.

Conclusions

In conclusion, incidental subclinical atherosclerosis on ED-performed CCTA is associated with increased likelihood of subsequent statin prescription and persistently lower LDL cholesterol levels. Cardiologist involvement in a patient’s care is also independently associated with greater likelihood of statin prescription after adjustment for cardiovascular risk. However, many people with both elevated ASCVD risk based on clinical risk factors and subclinical atherosclerosis on CCTA do not receive recommended preventive therapy. These findings highlight opportunities to improve primary preventive cardiovascular care from the increasing use of CCTA in the ED to evaluate chest pain.

Competency in Medical Knowledge.

The presence of non-obstructive coronary artery disease (CAD) is independently associated with future cardiovascular disease risk. The finding of non-obstructive CAD on coronary CT angiography performed in the Emergency Department is associated with a 7-fold increased likelihood of statin prescription compared to individuals without CAD, though many individuals who meet criteria for primary prevention statin therapy remain untreated even after CT coronary imaging reveals subclinical atherosclerosis. This gap highlights opportunities to improve the allocation of indicated cardiovascular preventive therapies.

Translational Outlook.

Strategies and systems to improve allocation of statin prescriptions and other cardiovascular preventive therapies after imaging reveals subclinical atherosclerosis (e.g., automated reporting and decision support for inpatient and longitudinal providers) deserve prospective study. Additionally, risk scores to integrate clinical characteristics and imaging findings, particularly when clinically estimated risk and imaging findings are discordant, warrant dedicated study.

Abbreviations

ASCVD

atherosclerotic cardiovascular disease

BMI

body mass index

CAC

coronary artery calcium

CAD

coronary artery disease

CAD-RADS

Coronary Artery Disease Reporting and Data System

CCTA

coronary CT angiography

ED

Emergency Department

LDL

low-density lipoprotein

PCE

Pooled Cohort Equations