Patient characteristics and acute cardiovascular event rates among patients with very high‐risk and non‐very high‐risk atherosclerotic cardiovascular disease

Abstract Background The risk for subsequent major cardiovascular (CV) events among patients with very high‐risk (VHR) atherosclerotic CV disease (ASCVD) remains to be fully elucidated. Hypothesis We assessed the characteristics and major CV event rates of patients with VHR versus non‐VHR ASCVD in a real‐world setting in the United States (US), hypothesizing that patients with VHR ASCVD would have higher CV event rates. Methods This was a retrospective cohort study conducted from January 01, 2011, to June 30, 2018, in the US using the Prognos LDL‐C database linked to the IQVIA PharMetrics Plus® database supplemented with the IQVIA prescription claims (Dx/LRx) databases. Patients were ≥18 years old and had  ≥2 non‐ancillary medical claims in the linked databases at least 30 days apart. The study was conducted in 2 stages: (1) identification of patients with ASCVD who met the definition of VHR ASCVD and a matched cohort of non‐VHR ASCVD patients using the incidence density sampling (IDS) approach; (2) estimation of the occurrence of major CV events. Results Among patients with ≥1 major ASCVD event (N=147,679), most qualified as VHR ASCVD (79.5%). There were 115,460 patients each in IDS‐matched VHR and non‐VHR ASCVD cohorts. The composite myocardial infarction/ischemic stroke event rates in the VHR and non‐VHR ASCVD cohorts were 8.04 (95% confidence interval [95% CI]: 7.87‐8.22) and 0.82 (95% CI: 0.77‐0.88) events per 100 patient‐years, respectively, during the 1‐year post‐index period. Conclusions Most patients with ≥1 previous major ASCVD event treated in real‐world US clinical practice qualified as VHR ASCVD. Patients with VHR ASCVD had much higher rates of major CV events versus non‐VHR ASCVD patients.

K E Y W O R D S atherosclerotic cardiovascular disease, major cardiovascular events, real-world evidence, very high-risk

| INTRODUCTION
Low-density lipoprotein cholesterol (LDL-C) is a modifiable causal risk factor in the pathogenesis of atherosclerotic cardiovascular (CV) disease (ASCVD), 1 with lower LDL-C levels associated with a reduced risk of CV events and improved patient outcomes. [2][3][4][5][6] Updates in the 2018 American College of Cardiology/American Heart Association (ACC/AHA) multi-society blood cholesterol guideline introduced the very high-risk (VHR) ASCVD category. 7 These patients have a history of multiple major ASCVD events (i.e., recent acute coronary syndrome [ACS] ≤12 months, history of myocardial infarction [MI] >12 months, history of ischemic stroke [IS], or symptomatic peripheral arterial disease [PAD]), or a single major ASCVD event and multiple high-risk conditions. 7 The ACC/AHA 2018 guideline recommends that all patients with VHR ASCVD receive lipid-lowering therapy (LLT) with high-intensity or maximally tolerated statin therapy. 7 For patients with VHR ASCVD with LDL-C ≥70 mg/dL (≥1.8 mmol/L) despite optimized statin therapy, the addition of ezetimibe and proprotein convertase subtilisin/kexin type 9 (PCSK9) inhibitors is recommended. 7 Considering the recent introduction of the VHR stratification, 7 the clinical characteristics, including treatment patterns and risk for subsequent major CV events, among patients with VHR versus those with non-VHR ASCVD remain to be fully elucidated in routine clinical practice. Real-world characterization of the VHR ASCVD population is important as these patients are likely to benefit from intensive LLT with the addition of non-statin therapies such as PCSK9 inhibitors. [4][5][6]8,9 Therefore, the current study had two objectives: first, to describe patient characteristics, utilization of LLT, and LDL-C levels among patients with ASCVD who met the definition of VHR per the 2018 ACC/AHA cholesterol guideline 7 versus patients with ASCVD not meeting the VHR criteria; and second, to estimate the rates of subsequent major CV events in VHR ASCVD and non-VHR ASCVD cohorts, with analyses by type of major ASCVD event.

| Study design and patients
This retrospective cohort study was conducted in the United States (US) using the Prognos LDL-C database (Prognos Health, New York, NY, USA) 10 linked to the IQVIA PharMetrics ® Plus database supplemented with the IQVIA prescription claims (Dx/LRx) databases (IQVIA, Plymouth Meeting, PA, USA). 11 The Prognos LDL-C database has been previously used in retrospective cohort studies. [12][13][14] The aggregated IQVIA PharMetrics Plus database comprises adjudicated claims for patients across the US and is sourced directly from insurance companies, and contains data on patient's health plan claims, demographics, clinical characteristics, and occurrence of CV events.
The IQVIA LRx database captures information on adjudicated dispensed prescriptions sourced from retail, mail, long-term care, and specialty pharmacies. The IQVIA Dx database contains unadjudicated medical claims from office-based physicians, ambulatory facilities, and hospital-based physicians, and is sourced from clearing houses involved in the claims processing. The IQVIA databases linked to the Prognos LDL-C database were accessible to the authors of this study, and a primary study dataset was constructed using linked patient data from the databases. The study was conducted in two stages ( Figure 1). In stage 1 of the study, patients with VHR ASCVD were identified by the presence of the 2018 ACC/AHA guideline criteria 7 for VHR ASCVD (i.e., 1 major ASCVD event and 2 risk factors, or ≥2 major ASCVD events) during the 5-year pre-index period (January 01, 2011, to December 31, 2015). The operational definitions used to identify the VHR ASCVD criteria are reported in Supplementary Table 2 To obtain comparable cohorts, patients with VHR ASCVD were matched to a cohort of non-VHR ASCVD patients with comparable patient demographics (age, sex, region) in a 1:1 ratio using the incidence density sampling (IDS) method, also referred to as the risk set sampling method. [15][16][17] The IDS methodology followed a "sampling with replacement approach," where a patient with non-VHR ASCVD could be matched to multiple patients with VHR ASCVD, and patients sampled in the non-VHR cohort were eligible to become patients with VHR ASCVD at a later date. The rationale for using the IDS methodol- To match a non-VHR ASCVD patient with a VHR ASCVD patient, we first identified all patients in the study cohort who were at risk on the index date (diagnosed with ASCVD, not lost to follow-up, and not VHR ASCVD)-referred to as the risk set. From the risk set, a patient with non-VHR ASCVD was matched to a patient with VHR ASCVD on age (±3 years), sex, and region using a greedy match algorithm. 18 The matched non-VHR ASCVD patient was then assigned an index date equivalent to the corresponding case. The process was repeated until all patients with non-VHR ASCVD were paired with those with VHR ASCVD, resulting in a matched cohort. Patients were excluded from the study if there were missing data (such as age and sex), and if matched non-VHR ASCVD patients could not be identified using the IDS method.
Patients with VHR ASCVD were further grouped into the following mutually exclusive subgroups in a hierarchical manner based on the major ASCVD event(s) that led to their qualification as VHR ASCVD: (1) patients with ≥2 major ASCVD events; (2) patients with 1 major ASCVD event, which was recent ACS; (3) patients with 1 major ASCVD event, which was a history of MI (non-recent ACS); (4) patients with 1 major ASCVD event, which was a history of IS; (5) patients with 1 major ASCVD event, which was symptomatic PAD.
In stage 2 of the study, the occurrence of major CV events in the IDS-matched VHR ASCVD and non-VHR ASCVD cohorts was esti-

| Outcomes
We assessed demographics at index (age, sex, insurance type, and geographical region), current LLT patterns in the 90 days and 1-year

| Ethics
This was a retrospective analysis of de-identified aggregate claims data; therefore, informed consent, ethics committee approval, or institutional review board approval was not required. The study complied with all applicable laws regarding patient privacy, using Health Insurance Portability and Accountability Act-compliant de-identified retrospective data sources. No direct patient contact or primary collection of individual human patient data occurred. Study results were in tabular form and aggregate analyses, which omitted patient identification information. All authors had full access to all the data in the study and take responsibility for its integrity and data analysis.

| Statistical analysis
Analyses were conducted using SAS version 9. Mortality data were unavailable in the linked databases and were not included in the calculation of major CV event rates.

| Baseline demographic and clinical characteristics
Baseline demographic and clinical characteristics are reported in

| Overall CV event rates post-index
In the VHR ASCVD cohort, the median (interquartile range  ( Figure 2; Table 3). Major CV event rates defined by the VHR ASCVD subgroups in the 1-year post-index period are summarized in Table 3  patient-years, respectively). 24 Of note, the major CV event rates observed in the current study were higher than those in the analysis of the MarketScan health insurance database. 24 This may be accounted for by differences in methodology as in the current study, patients were followed-up immediately after they qualified as VHR ASCVD; thus, there was higher CV risk captured in the current analysis than in the previous analysis of the MarketScan health insurance database. 24 Broadly, this study demonstrates that patients with ASCVD were exposed to a high residual CV risk due to suboptimally controlled LDL-C above ACC/AHA 2018 guideline recommendations, 7  has the potential to reduce major CV event rates and improve patient outcomes. [4][5][6]8 Indeed, it is now recognized that patients with the highest levels of CV risk derive greater absolute and relative risk reductions with the addition of non-statin therapies (e.g., ezetimibe and PCSK9 inhibitors) than those with lower CV risk. 6,8,9,25 Moreover, the value of PCSK9 inhibitors is improved by selecting patients at higher risk for the occurrence of CV events, 26 and the major CV event rates observed in the current study were within the range where the addition of PCSK9 inhibitors to background LLT would meet cost-effectiveness thresholds from models based on ACC/AHA guidelines. 7,26 The results of this study should be considered within the context of several limitations. First, it was a retrospective analysis using linked commercial claims databases and was therefore subject to the inherent limitations of this methodology, including limited generalizability to patients without commercial insurance and to those aged ≥65 years. Second, there was a lack of CV mortality data, a limitation well-recognized for US-based claims datasets, and this information was therefore not included in the assessment of major CV events.
However, this limitation would be expected to result in an underestimation, rather than an overestimation, of CV events during follow-up. in meaningful differences between matched patients with ASCVD being compared, beyond the VHR categorization.
In conclusion, the majority of patients with ≥1 previous major ASCVD event qualified as VHR ASCVD in real-world US clinical practice.
In patients with VHR ASCVD, LLT utilization rates were relatively low and LDL-C was suboptimally controlled, even in patients receiving highintensity statins and/or ezetimibe treatment. Application of the ACC/AHA 2018 guideline VHR ASCVD criteria was able to identify patients with higher rates of overall major CV events (compared with those not meeting the VHR ASCVD criteria), who would therefore derive the greatest absolute benefit from more intensive LDL-C lowering.

ACKNOWLEDGMENTS
Medical writing support, which was in accordance with Good Publica-

DATA AVAILABILITY STATEMENT
Qualified researchers may request data from Amgen clinical studies.
Complete details are available at the following: http://www.amgen. com/datasharing.